Bio-based and Applied Economics 8(1): 21-61, 2019 ISSN 2280-6180 (print) © Firenze University Press ISSN 2280-6172 (online) www.fupress.com/bae Full Research Article DOI: 10.13128/bae-8145 The impact of assistance on poverty and food security in a fragile and protracted-crisis context: the case of West Bank and Gaza Strip Donato Romanoa, Gianluca Stefania,*, BeneDetto Rocchia, ciRo fioRillob a Department of Economics and Management, University of Florence, Italy b Food and Agriculture Organization of the United Nations, West Bank and Gaza Strip Office, Jerusalem1 Abstract. This paper assesses the impact of assistance on the wellbeing of Palestinian households. The impact evaluation analysis uses a difference-in-difference approach for the treatment of sample selection bias. The paper uses data from the 2013 and 2014 rounds of the Palestinian Socio-Economic and Food Security survey to estimate the impact of assistance to West Bank and Gaza Strip (WBGS) households on their poverty and food security status. Results suggest that both poverty and food inse- curity would have been much higher for WBGS as a whole without assistance, fur- ther increasing in areas with lower levels of assistance. However, the average positive impact of assistance hides a lot of heterogeneity. In fact, while there is a clear posi- tive impact of the intensity of assistance on poverty reduction, food consumption and diet diversity in the West Bank, Gaza Strip analysis shows mixed results. Results high- light how the international community cannot disengage from supporting Palestinian households without severely impacting their wellbeing. Keywords. Poverty, food security, impact analysis, West Bank and Gaza Strip. JEL codes. Q18, I32. 1. Introduction The relationship between foreign assistance2 and development is one of the most debated topics in development policy. Since World War II, the debate shifted from dis- cussing the rationale for mobilizing foreign resources to boost economic growth (Chenery and Bruno, 1962; Chenery and Strout, 1966; Lal, 1972), to assessing the impact of aid on 1 The views expressed in this paper are those of the author(s) and do not necessarily reflect the views of the Food and Agriculture Organization of the United Nations. 2 Foreign assistance is a broad term for any voluntary transfer of resources from one government, international organization, or NGO to a recipient country, usually a developing country. It encompasses loans (both soft or hard) and grants as well as in-kind transfers and technical assistance. The paper uses “foreign assistance” inter- changeably with the term “foreign aid”. *Corresponding author: gianluca.stefani@unifi.it 22 D. Romano et alii economic growth and poverty reduction (Bauer and Yamey, 1982; Cassens & Ass., 1986; Krueger, 1986; Mosley, 1987; Collier and Dollar, 2001 and 2002), and subsequently to generating evidence in order to better design interventions and enhance aid effectiveness (Burnside and Dollar, 2000; Hansen and Tarp, 2001). More recently, increasing attention has been devoted to assessing the effectiveness of assistance delivered in fragile contexts.3 This shift was driven by empirical evidence sug- gesting that natural, economic and political risks are rising across the world (World Bank, 2011; Zseleczky and Yosef, 2014), as well as by the rapidly growing literature on fragile states (Ipke, 2007; Kaplan, 2008; Zoellick, 2008; Baliamoune-Lutz and McGillivray, 2008; Stewart and Brown, 2009; Andrimihaja et al., 2011; Chandy, 2011; Naudé et al., 2011). The key question here is whether aid can deliver its expected results within fragile/conflict contexts. The literature shows mixed empirical evidence (Dollar and Levin, 2006; Fielding and Mavrotas, 2008; Ishihara, 2012; Chandy et al., 2016). As a result, many practitioners, policymakers, and even laypeople express mounting concern for the poor development records within fragile country contexts. This implies a need to develop new approaches that explicitly address fragility pathways to insecurity when designing development and humanitarian assistance strategies in fragile/conflict contexts (OECD, 2007; EU Commis- sion, 2009; World Bank, 2011).4 This paper contributes to the empirical literature on the impact of assistance in frag- ile contexts by adopting a microeconomic perspective. It aims to estimate the impact of assistance intensity on household wellbeing proxied by two outcome dimensions, poverty and food security. We adopt a counterfactual framework using a difference-in-difference approach to address sample selection bias as well as instrument variable econometric modeling to get rid of endogeneity problems where appropriate (e.g. poverty reduction). The empirical application focuses on the specific fragile, protracted-crisis context of the West Bank and Gaza Strip (WBGS) as a case study. This specific region was chosen for several reasons. WBGS has been among the highest per-capita recipient of official development assistance worldwide (World Bank, 2019) and it is also experiencing one of the longest contemporary conflict in the world. The majority of Palestinians living under occupation would be unable to meet their own bare necessities since both humanitar- ian and development interventions in WBGS are largely financed by foreign assistance. Indeed, the pledge for humanitarian assistance, amounting to USD 540 million in 2018 (OCHA, 2017a), is completely financed by foreign resources. Similarly, the share of for- 3 There is no universally accepted definition of fragility. Instead of trying to stringently define fragility, OECD (2015) identifies fragile contexts according to a multi-dimensional framework that helps reveal different patterns of vulnerability in a given country. The five fragility “clusters” considered by OECD are the following: widespread violence, limited justice, ineffective and unaccountable institutions, weak economic foundations, and low resil- ience to shocks and stressors. These characteristics substantially impair the fragile country’s economic perfor- mance, the delivery of basic social services, and the efficacy of donor assistance. 4 This was explicitly considered by the so-called “New Deal for Engagement in Fragile States” announced at Busan in 2011. This deal identified five “Peace-building and State-building Goals”: legitimate politics, security, justice, economic foundations, and revenues and services (https://www.pbsbdialogue.org/en/). It also considered in the United Nation’s “New Way of Working”, (https://www.un.org/jsc/content/new-way-working) within which humanitarian, development and peace actors are called to work together to pursue collective outcomes over multiple years to overcome the traditional divide between humanitarian and development interventions. This is at the core of the so-called “Triple nexus”, which aims to integrate the humanitarian, development and peace aspects of interventions. 23The impact of assistance on poverty and food security in a fragile and protracted-crisis context eign support in 2018 accounted for as much as 48% of total development expenditure, a sum roughly equal to USD 381 million (IMF, 2018). Between 2007-2016, the yearly average total of aid received amounted to more than 2.3 billion USD per year or 23% of Palestinian GDP (WDI, 2018). Despite such large aid inflows, the Palestinian GNI per capita is still around USD 3,180 (WDI, 2018), qualifying WBGS as a lower-middle income country. Similarly, the Palestinian HDI is 0.686, placing WBGS 119th out of 189 countries and territories (UNDP, 2018). According to the humani- tarian needs assessment (OCHA, 2018), some 2.5 million people are in need of assistance on a total population of 4.95 million and 1.9 million people are targeted by humanitarian interventions. In factIn light of this, data released by the Palestinian Central Bureau of Statistics (PCBS) regarding the 2013 and 2014 Socio-Economic and Food Security (SEF- Sec) survey data (FSS-PCBS, 2016) — designed for the first time as a panel — offers a unique opportunity to assess the impact of assistance on household poverty and food security in WBGS. It is important to note that from 2013 to 2014, the period in which the data was collected, the region faced persistent occupation as well as an open-arm conflict in the Gaza Strip occurring from July 2014 to August 2014. To conduct the aforementioned analysis, the paper is organized in the following way: Section 2 reviews the literature on aid and development, looking at both theoretical argu- ments and empirical results. Section 3 introduces the Palestinian context and provides an overview of assistance to Palestinian households. Section 4 analyzes Palestinian house- holds’ profiles in terms of poverty and food security at the beginning of the period of analysis. Section 5 describes the data and methods used in the impact evaluation. Section 6 discusses the results of the impact evaluation analysis. Finally, Section 7 summarizes main findings and discusses policy implications. 2. Foreign Assistance and Development: An Introduction Foreign assistance can be traced back to the colonial period. At the time, European pow- ers provided large amounts of money to their colonies, typically to improve infrastructure, with the ultimate goal of increasing economic output. The use of foreign assistance as it is known today — as an instrument to help poor countries improve living standards — came into existence only after World War II (Thorbecke, 2000). The emergence of a new economic order and the founding of international organizations (such as the United Nations, the IMF, and the World Bank) following WWII shaped foreign aid to become what it is today. The success of the Marshall Plan, the US-sponsored package implemented between 1948 and 1953 to rehabilitate the economies of Western and Southern European countries, showed that capi- tal transfers alongside technical assistance could effectively spur growth so that targeted econ- omies were able to surpass their pre-war economic levels by 1952. Aid to developing countries today is more complex. Its use is determined by several intertwined motives, including altruism, access to markets and resources, geopolitics, and colonial legacies. The impact of foreign assistance to developing countries is mixed, with success stories in various South East Asian countries but also numerous failures in Sub- Saharan countries (Kanbur, 2000). Foreign aid is thought to have helped poor countries raise income per-capita growth rates, in some cases converging with high-income countries, successfully lifting large seg- 24 D. Romano et alii ments of the population out of poverty. However, this is difficult to establish unequivo- cally. There are two major difficulties when analyzing the relationship between foreign assistance and development. Firstly, there are issues with different theoretical frameworks– macro as well as micro–that provide the rationale for foreign aid interventions. Secondly, empirical studies lack conclusive evidence, making it hard to identify causal links between aid and development outcomes. Indeed, there is a large gap between aid achievements at the micro and macro levels, with greater difficulties in establishing causalities at the mac- ro/country level compared to the micro/project level. This is the so-called “micro-macro paradox” (Mosley, 1987). As a consequence, the effectiveness of aid in the promotion of development is often uncertain and controversial, with personal opinions often deeply founded in ideology. The consequence is an ongoing debate regarding best practices in the provision of for- eign assistance aptly named the “aid debate”.5 Positions on the matter range from strong believers in the potential effectiveness of foreign aid who advocate for even more aid (Sachs, 2005), to deep skeptics stressing the importance of experimentation and learn- ing from past mistakes (Easterly, 2006). Along this spectrum lie pragmatists who sup- port peace and the use of a broad set of instruments (Collier, 2007) as well as opponents endorsing anti-corruption practices to increase aid effectiveness (Moyo, 2009). Finally, the aid debate also includes scholars who argue for the reduction of damaging OECD trade policies in agriculture, increased provision of technical assistance regarding institution building, an increase in investment devoted to fighting diseases and improving agricultur- al technology in tropical environments, and greater support for institutional reforms that favor secure property rights, the rule of law, and a reduction in arms sales to developing countries (Deaton, 2013). 2.1 Macroeconomic perspective The most important theoretical arguments in support of foreign aid as an effective strategy to boost growth and catch-up to rich countries are rooted in Keynesian growth models. The Harrod-Domar model (Domar, 1957) provides theoretical background for growth in developing country contexts by identifying savings rate and choice of technique (via the incremental capital-output ratio, or ICOR) as the two determinants of a country’s growth rate. The policy implications the model suggests for accelerating growth are clear: raise the savings rate (i.e. promote savings through stronger financial institutions) and lower the ICOR (i.e. increase the marginal productivity of capital through better technol- ogy). This is where foreign aid transfers come in. Using these transfers for investment can fill domestic saving gaps in developing countries, thus providing a “Big Push” to kick-off economic growth (Rosenstein-Rodan, 1943). A popular extension of the Harrod-Domar model devised in the 1960s in Latin America defined two kinds of capital goods used in production. The first kind are capital goods of domestic origin, such as buildings financed by domestic savings, while the other 5 Foreign aid has throughout its history been subjected to close scrutiny both by academic researchers and oth- ers (Dethier, 2008). A large literature extending over several decades bears witness to this, and the boundary between policy advocacy and research has not always been clearly delineated. 25The impact of assistance on poverty and food security in a fragile and protracted-crisis context kind are of foreign origin, such as imported intermediate goods and machinery paid for using foreign savings. If the two forms of capital are in fixed proportion, then the scarcest of the two types of savings will always be binding. This is the core of the “two-gap model” (Chenery and Bruno, 1962; Chenery and Strout, 1966). Foreign aid that increases foreign savings can effectively increase growth with enough domestic savings, despite a deficit of foreign exchange. However, foreign aid cannot translate to growth if there is a deficit of domestic savings even if an economy has enough foreign exchange to acquire necessary amounts of imported capital goods. Subsequent developments are based on new growth theory, which endogenously explains productivity growth by extending the above paradigm with an analytical basis for empirical cross-country studies (Robinson and Tarp, 2000). The underlying causal chain runs from aid to savings, from savings to investment, and finally from investment to growth. In the new growth theory approach, investment and productivity variables are assumed to depend on policy and institutional variables. Usually, the effectiveness of aid has been empirically tested using country-level macro data, with aggregated aid as a single resource. Such tests examined whether more aid lead to better outcomes, in particular whether more aid lead to higher growth. It is no surprise that reduced-form analysis shows tenuous links between aid and development outcomes, since aid is often advanced for non-developmental objectives, such as disaster relief or military and political ends. As emphasized by Bourguignon and Sundberg (2007: 317) development economists must better understand “the links from aid to final outcomes” because “trying to relate donor inputs and development outcomes directly, as through some kind of black box, will most often lead nowhere.” Opening the black box allows for the identification of three types of links–from donors to policy-makers, from policy- makers to policies, and from policies to outcomes–which, in turn, may provide additional answers. Empirical studies on the link from donors to policymakers reveal a body of circum- stantial evidence built primarily on years of failed aid efforts (Dollar and Levin, 2006). Donor views regarding the “right development policies” have been promoted through aid conditionality with little attention to specific country contexts. For instance, public enter- prise privatization and finance liberalization have at times been regarded as necessities, though were encouraged with little regards for local socioeconomic conditions, making such measures ineffective, risky, or simply counterproductive. The link from policymak- ing to policy formulation and implementation depends largely on governance systems. There is evidence suggesting the association between good governance and good poli- cies, although the direction of causality is hard to determine. In practice, most research has focused on the relationship between governance and development outcomes, bypass- ing the impact on policies and pointing instead to the importance of good governance for better outcomes (Acemoglu et al., 2005). Regarding the impact of policies on outcomes, there is a good understanding of the effect of macro stability, investment climate, as well as well-managed trade openness on growth, even though country specificity can make it hard to generalize the impact of these factors. Cross-country comparisons however indi- cate that better-quality policies are associated, on average, with higher GDP growth. Some authors use empirical analyses to argue that aid leads to growth with decreas- ing returns (Hansen and Tarp 2001). Others suggest that national growth-inducing pol- 26 D. Romano et alii icies may reduce aid effectiveness because good policies and aid are substitutes of each other (Dalgaard and Hansen 2001). Finally, some authors hold that aid stimulates growth conditional on key features. For instance, it is often argued that aid works if provided to countries that implement good policies (Burnside and Dollar 2000). This conclusion was questioned by Easterly et al. (2004) who showed that the aid-policy relation was not robust enough for the expansion of the database in years and countries. Despite such dif- fering positions, cross-country regression analysis largely concludes that the relationship between aid and development outcomes is weak and often ambiguous (Rajan and Subra- manian, 2005; Clemens et al., 2004). In recent years, econometric assessments have included meta-analyses to synthesize results from the existing body of empirical data while controlling for heterogeneity among studies. Surprisingly, even these studies, which are supposed to provide more objec- tive analyses, have contributed little to resolving the aforementioned controversies. Con- sider, for instance, two such studies by Doucouliagos and Paldman (2009) and Mekasha and Tarp (2013): while the former failed to find any significant impact of foreign aid on growth, the latter found an impact that is both positive and statistically significant. In conclusion, macro growth effects are both harder to achieve and harder to observe. They are harder to achieve than micro growth effects because the magnitude of aid may not be sufficient to affect recipient countries’ macro variables, and harder to observe because causality is difficult to establish in cross-country regressions (Mavrotas, 2015). 2.2 Microeconomic perspective Non-conclusive results of reduced-form cross-country aid regressions brought about the need to establish the channels through which aid mattered the most for economic growth and poverty reduction (Dalgaard et al., 2004). This was done through empirical studies at the micro level that analyzed the impact of single project and program inter- ventions. Until the 1990s, these evaluations were based on cost-benefit analysis (CBA) of single projects by computating the internal rate of return of the intervention. Such studies show that aid is effective at the micro level when taking into considerations local projects (Hirschman, 1967; Mehrotra and Jolly, 1997). However, these results came under severe criticism once the concept of aid fungibility, i.e. aid money being used for purposes other than those earned, spread. In fact, rate-of-return metrics ignore more complex opportu- nity-cost issues like the fungible use of foreign aid. The approach also became problem- atic as donors started to embrace broader goals for aid, such as environmental sustain- ability and multiple social goals with hard-to-quantify objectives. In parallel, the weakness of CBA-based impact evaluations, summarized under headings such as “before-and-after” and “with-and-without,” was the topic of many debates. Consequently, methodological issues became increasingly important in the aid-effectiveness debate (Cassen & Ass., 1987; World Bank, 1998). More recently, knowledge at the micro and project level has expanded based on evalu- ations using advanced econometric techniques and rigorous experimental or quasi-exper- imental designs. Econometric techniques are used to examine the impact of specific poli- cies or projects on local communities, household decision making, and individual welfare (Banerjee and Duflo, 2011). Given the number projects and their different impacts in var- 27The impact of assistance on poverty and food security in a fragile and protracted-crisis context ying country circumstances, continued evaluation and revision is needed. Impact evalua- tion evidence began in the mid-1990s. By the turn of the century, impact evaluation pub- lications became increasingly more common, continuing to date (Cameron et al., 2016). Rigorous ex-post impact evaluations help inform government and donor decisions, an idea supported by donor agencies and even by aid critics (e.g., Easterly 2006). However, an evaluation gap still exists. This is because governments, official donors, and other funders do not demand or produce enough impact evaluations and those that are conducted are quite often methodologically flawed (Savedoff et al., 2006). This calls for a system- atic review of conclusions drawn from such studies. Several initiatives have been imple- mented in response to this issue, such that many reviews and meta-analyses are in circu- lation today. In terms of sectors, the ones most represented in studies are social protec- tion, health and nutrition, and education. Cash transfers is the most represented modality, though in-kind transfers and vouchers are also well-researched, especially in the context of humanitarian crises. Randomized control trials and difference-in-difference studies are the most widely used methods. Studies assessing the causal relationship between interventions and outcomes of humanitarian assistance generally lack a reliable and robust base of evidence (Clarke et al., 2014). Only a small proportion of the many evaluations of humanitarian assistance use designs with a counterfactual, control or comparator group that allows the studies to attribute measurable changes outcome indicators to programs or policies. However, there are also several examples of randomized trials. It is possible to generate evidence for spe- cific questions using randomized trials, although this evidence base is limited and concen- trated in certain areas, such as mental health (Cameron et al., 2015). Foreign aid has generally brought about a positive contribution in education, the most tangible outcome being increased enrolment rates in primary education (Riddell and Niño-Zarazúa, 2016; Birchler and Michaelowa, 2016). However, there is a consider- able gap regarding the contribution of aid to improvements in the quality of education. Masino and Niño-Zarazúa (2016) conducted a systematic review of experimental and quasi-experimental evidence to establish what works best to improve education quality in developing countries. They found that educational policies are most successful when implemented in combination with multiple interventions. Aid channeled into a variety of interventions, targeting different educational levels and utilizing different aid modalities works best. Considering this heterogeneity, it should not be surprising that a generalized blueprint applicable to all developing countries hasn’t been devised. Literature in the food security and nutrition sector has a lot of variation in program implementation (e.g. size and modality of transfers, duration and frequency of transfers, strength of conditions, pre-existing levels of undernutrition, health services). This makes difficult to establish which of the various interventions on food security and nutrition is most effective. Conclusions of summary studies range from cautiously optimistic (Ahmed et al., 2009; Ruel et al., 2013) to lacking significant results (Manley et al., 2012). In 2016, Doocy and Tappis reviewed 108 studies on intervention modalities. They found that unconditional cash transfers and vouchers may improve household food security among conflict-affected populations and maintain household food security during crises specifi- cally affecting food, such as droughts. Moreover, unconditional cash transfers led to great- er improvements in dietary diversity and quality than food transfers. Food transfers were 28 D. Romano et alii found to be more effective in increasing per capita caloric intake than unconditional cash transfers and vouchers. While the evidence reviewed offers some insights, the scarcity of rigorous research on cash-based approaches limits the strength of such findings. However, drawing on findings from randomized control trials, Karlan and Appel (2012) identify seven ideas that work: microsavings; reminders to save; prepaid fertilizer sales; deworming; remedial education in small groups; chlorine dispensers for clean water; and commitment devices. Likewise, Banerjee and Duflo (2011) draw on experimental studies to identify a host of promising interventions in areas ranging from health and education to policing. Though promising, impact evaluation studies have several limitations. It is illusory to believe that all interventions can be subject to impact evaluations and that such evalua- tions will permit the flow of aid exclusively to what works, as some have suggested (East- erly, 2006; Banerjee, 2007). It is impossible to evaluate all projects. Evaluations can also be misleading when projects or programs are applied outside the context in which they were evaluated, meaning there is a serious problem of external validity (Pritchett and San- defur, 2013). Furthermore, many policies have general equilibrium effects often ignored by impact evaluations. This suggests that unlocking the secret of aid effectiveness is most likely to be revealed by trial and error than by randomized control trials (Deaton 2013). Nonetheless, experimental and quasi-experimental studies are grossly underutilized instruments with tremendous scope to improve and regularize their use in bilateral and multilateral donor agencies. A larger evidence base and a more standardized approach to documenting and comparing costs and benefits of interventions are needed to draw important conclusions on the effectiveness of different development interventions (Save- doff et al., 2006; White, 2010; Cameron et al. 2016). 2.3 Aid effectiveness in fragile contexts The new economics of aid stresses the importance of good governance to successfully achieve growth. Focusing on good governance leads to country selectivity such that trans- fers are targeted at countries that pass the good-policy test. This means aid is shifted from project financing to budget financing. However, targeting countries with high institutional and policy scores means that poor individuals in countries with failed states and in post- conflict societies will not be reached. The problem of building a developmental state that qualifies for aid is also left open. Social development funds, local governments, and NGOs therefore play an important role: they can bypass central governments while capacity building for improved governance goes on. Traditional empirical research has largely dismissed the analysis of fragile or conflict contexts. For instance, econometric evidence used in the aid-effectiveness debate suggests that the ineffectiveness of aid is due to the failure of the recipient governments to create the right policy environment. However, this data uses a cross-section of countries with- out any specific focus on fragility contexts that, at best, were treated as a dummy variable in the regressions (Boone, 1995; Burnside and Dollar, 2000; Hansen and Tarp, 2001; Dal- gaard and Hansen, 2001; Easterly et al., 2004; Doucouliagos and Paldam, 2009). The same reduced-form approach based on country aggregate data has been adopted in more recent 29The impact of assistance on poverty and food security in a fragile and protracted-crisis context literature on the “growth-efficient” level of aid.6 The literature found that the relationship between aid and growth takes on an inverted-U shape for both fragile and non-fragile countries, identifying a lower growth-efficient level of aid in the former as compared to the latter (Gomanee et al., 2005; McGillivray et al., 2006; McGillivray and Feeny, 2008; Feeny and McGillivray, 2009; Naudé et al., 2011). Existing evidence from impact evaluations in fragile contexts is equally poorly devel- oped. A recent evidence gap map review of impact evaluations found little to no evidence on most categories related to the five Peace-building and State-building Goals. Only two Goals (community-driven reconstruction and psycho-social programs for victims) had a large enough number of studies to be promising for evidence synthesis. While prioritiz- ing new research in understudied areas might help fill such knowledge gaps, the nature of experiments also imposes limits on what is studied. In addition to the common limita- tions of randomized studies (cf. section 2.2), some interventions may be impractical or unethical in fragile/conflict contexts (Humphreys, 2015). Some authors therefore look beyond the standard impact evaluation approach, choosing instead to focus on the drivers of success in fragile contexts by developing comprehensive theories that identify impor- tant factors and establish how they interact to create outcomes. The authors then test or demonstrate the plausibility of their arguments through case studies (cf., for example, Guisselquist, 2015). Addison (2000) was one of the first in the field to discuss the role of aid before, dur- ing, and after armed conflicts. He found that aid distributed during conflicts plays a minor yet positive role in humanitarian assistance as well as in the transition from war to peace. There are, however, serious problems in operating in wartime environments. This author notes that aid can complicate conflicts when it falls into the hands of belligerents. After periods of war, aid plays a major role in rehabilitation and reconstruction efforts. Finally, Addison considers the possibility of using aid to prevent conflict in areas at risk, arguing that foreign policy support should incorporate aid in conflict prevention efforts. Such aid should focus on reducing poverty and inequality to dampen social tensions as well as sup- port institutions and processes for conflict resolution. More recently, Guisselquist (2015) argued that development assistance to fragile states and conflict areas can act as a core component of peacebuilding by providing support for the restoration of government functions, the delivery of basic services, the rule of law and economic revitalization. Significant gaps exist regarding what has worked, why it has worked and the transferability and scalability of such findings. Nevertheless, three broad factors can identify why some interventions work better than others. The first is the area of intervention and the related degree of engagement with local state institutions. The sec- ond factor relates to local contextual elements, including windows of opportunity, capacity and the existence of local supporters. Finally, the third set of factors deals with project or program design and management. While the third set of factors is largely transferrable and scalable, the first two are less so and should be considered carefully when assessing the feasibility of extending project or program models to new contexts. Area of interven- tion, degree of engagement with domestic institutions and local contextual elements are 6 The so-called “growth-efficient” level of aid is the level of aid beyond which more aid is associated with lower growth. 30 D. Romano et alii vital factors to consider when making adjustments to improve the viability of development programs. Finally, a more radical approach was proposed by authors adopting a political econo- my perspective to analyze the workings of aid in conflict contexts (Murshed, 2002; Sogge, 2002; Kanafani and Al-Botmeh, 2008; Hever, 2010; Taghdisi-Rad, 2011 and 2015). The authors argue that the debate on aid effectiveness in fragile contexts has treated conflict as an external factor to be considered only at a much later stage in the analysis. They believe that a conflict and its interaction with local socio-economic structures should instead be the starting point of the analysis. As Taghdisi-Rad (2015: 5) said, it is impera- tive to understand “the nature of [a] conflict and the ideological forces behind its con- tinuation … to construct a framework for the analysis of economic performance under any given conflict”. 3. Assistance to Palestinian Households 3.1 West Bank and Gaza Strip: a fragile and protracted crisis context The world’s longest on-going crisis is in the West Bank and Gaza Strip, marked by more than fifty years of occupation, repeated waves of violence, and wars. The last two decades of Palestinian history have been marked by the construction of a separation bar- rier, the closure of the Gaza Strip in 2007, three devastating conflicts in 2008/2009, 2012 and 2014 respectively, as well as the increasing territorial fragmentation resulting from the continued expansion of Israeli settlements in the West Bank. The hope for greater welfare and stable economic growth brought about by the Oslo Accords (1993-95) has withered as a result of the unresolved Israeli-Palestinian conflict.7 Moreover, a growing political divide between the West Bank and Gaza Strip has further destabilized the economy. The attainment of Palestinian economic development is largely dependent on eco- nomic relations with Israel. According to the Paris Protocol, the Palestinian economy works under the framework of a customs and monetary union with Israel (Hever, 2015; UNCTAD, 2015).8 The Palestinian government cannot exert power over its borders nor 7 The Oslo Peace Accords, under which the Palestinian Authority (PA) was created in 1994, were intended to lead to a final negotiated settlement between the parties. The accords led to several administrative and security arrangements for different parts of the West Bank, which became divided in Areas A, B and C, with the PA having civil and security authority only in Area A (which accounts for 18% of the West Bank) and no author- ity whatsoever in Jerusalem. These were meant to be provisional terms, pending a final negotiated settlement. Permanent issues such as the status of Jerusalem, security arrangements, international borders, and the rights of Palestine refugees (5 million Palestine refugees are to this day dispersed across the Middle East) were left to be resolved after a five year interim period that ended in 1999. Twenty-five years after the Oslo Accords, no pro- gress has been made to settle the aforementioned pending issues (EU Commission, 2018). 8 The Protocol on Economic Relations, also called the Paris Protocol, is an agreement between Israel and the Palestine Liberation Organization signed in April 1994. It was incorporated into the Oslo II Accord of September 1995 with minor emendations. Originally, the Paris Protocol was to remain in force for an interim period of five years, yet it is still being enforced today. Essentially, the Protocol integrated the Palestinian economy into the Israeli economy through a customs union where Israel controls both Israeli and Palestinian borders (Elkhafif et al., 2014). The Protocol regulates the relationship and interaction between Israel and the Palestinian Authority in six major areas, namely: customs, taxes, labour, agriculture, industry and tourism. 31The impact of assistance on poverty and food security in a fragile and protracted-crisis context does it have an independent monetary policy.9 Economic growth suffers as a result of restrictions and controls placed on the movement of people and goods, access to resources such as land and water and access to productive inputs and markets. The Palestinian gov- ernment has limited ability of collecting its own taxes, while Israel recurrently withholds revenues collected on behalf of the Palestinians. Consequently Palestinian public financ- es are seriously destabilized. The situation is further complicated by the internal political divide that further limits the sovereignty of the Palestinian government. In such a situa- tion, the scope and geographical coverage of policy interventions has limited effectiveness. As long as barriers to trade, access, and movement remain high, the Palestinian econ- omy will continue on its current path of low growth.10 The Palestinian economy grew on average 5.5% per year over the last decade, with a marked difference between the West Bank and Gaza Strip. The economy slowed down in the last few years, so much so that 2017 estimates project GDP to fall from 3.1% to 1.7% per year in the medium-run (IMF, 2018). This is mostly due to the reduction of donor flows and the possibility of running tensions increasing further. With an expected population growth as high as 2.8% in 2017, the aforementioned implies a stagnation, if not a contraction, of per-capita incomes. Unemployment continues to be high (27.8% in 2017) and labor force participation con- tinues to be low, with structural unemployment particularly affecting young people and women: only 41% of youth between 15 and 29 years of age are active in the labor mar- ket while only 19% of women are active. Household and government consumption are the main drivers of economic activity. The two crowd out the investment necessary for faster growth. Primary capital inflows into Palestine are remittances and development assistance rather than FDI. Meanwhile, the national economy is highly import-dependent, Israel remaining by far its main trading partner. Overall, the Palestinian economy is still highly aid-dependent despite a sharp decline in aid. UNCTAD (2018) found that international developmental support to Palestine in 2017 amounted to USD 720 million, only one third of the USD 2 billion received in in 2008. Over the same period, budget support shrank from USD 1.8 billion to USD 544 million, a 70% decrease.11 Moreover, the fiscal burden of humanitarian crises and occu- pation-related fiscal losses have diverted donor aid from development to humanitarian interventions and budget support. As emphasized by UNCTAD (2015), no amount of aid would have been sufficient to put any economy on a path of sustainable development under conditions of frequent military escalations. Poverty and low standards of living are increasing in Palestine. The poverty headcount ratio at the national poverty line was estimated to be 29.2% in 2017 (PCBS, 2018a), well above the 2011 poverty headcount ratio of 25.8%. The proportion of poor in 2017 stood at 13.9% in the West Bank and 53.0% in the Gaza Strip. In that year, about 16.8% of Pal- estinians lived in extreme poverty (almost four percentage points more than in 2011), 9 The agreement defined specific arrangements through which the Government of Israel collects VAT, import duties and other so-called clearance (custom) revenues on behalf of the PA, sharing it with the latter on a monthly basis. These revenues account for 73% of the PA’s total net revenues (EU Commission, 2018). 10 World Bank (2017) estimates indicate that removing Israeli restrictions could increase annual GDP growth up to 10%. 11 The recent decision made by the United States to halt financial assistance to the Palestinian government and to UNRWA compounds an already critical situation. 32 D. Romano et alii with 5.8% residing in the West Bank and 33.8% in the Gaza Strip. The increase in overall poverty percentages between 2011 and 2017 is explained by the combined effect of two diverging dynamics: standards of living dramatically worsened in Gaza Strip, causing a rise in the poverty rate of 15 percentage points while poverty decreased by four percent- age points in the West Bank. According to the United Nations Relief and Works Agency for Palestine Refugees in the Near East (UNRWA), four out of five people living in Gaza’s are currently aid-dependent. Food and nutrition security are closely related to poverty. According to the Socio- Economic and Food Security Survey (FSS-PCBS, 2016), in 2014, 26.8% of total house- holds were classified as severely or moderately food insecure12. According to the Food Insecurity Experience Scale (FAO-IFAD-UNICEF-WFP-WHO, 2017), the prevalence of moderate or severe food insecurity in the population was 29.9% between 2014-16, of which 9.5% represented severe food insecurity. Stunting (or height-for-age) stood at 7.4% for children under the age of five in 2014-2016, while the prevalence of wasting (or weight-for height) was only 1.2%. Palestinians also face malnutrition: the prevalence of overweight youth was 8.2% among children under 5 years of age in 2014-2016 (FAO- IFAD-UNICEF-WFP-WHO, 2017). Micronutrient deficiency is also a concern among vul- nerable population groups, such as pregnant or lactating women and children. 3.2 An overview of assistance modalities in the West Bank and Gaza Strip Palestinians are vulnerable to many risks. According to OCHA (2018), the most criti- cal ones are the following: (i) the risk of conflict and violence, forcible displacement, and the denial of access to natural resources, inputs and markets that affect 2 million people in need of protection assistance; (ii) risks associated with poor water quality, poor waste- water collection and treatment, and lack of proper hygiene practices that affect 1.9 million people; (iii) the risks of food insecurity faced by 1.7 million people; and (iv) 1.2 million people are exposed to health and nutrition risks (e.g. conflict-related trauma casualties, pregnant and lactating women, children under the age of five, people with disability and elderly, etc.). Although all Palestinians are negatively impacted by the conflict, some of them – such as 1.4 million refugees, the 1.6 million Gazan civilians in need, and 0.4 mil- lion individuals living in Area C – are more severely affected (UNSCO, 2016). In the face of economic de-development and the denial of autonomous development prospects, humanitarian and development actors increasingly recognize the importance of bridging the humanitarian-development divide in Palestine. The result is a combination of emergency response measures with longer-term interventions to better address the causes of vulnerabilities faced by the Palestinian population (Diakonia, 2018).13 Many vulnerable 12 Preliminary results of the last SEFSec (PCBS, 2018b) show that the share of households classified as severely or moderately food insecure has increased by 6.2% between 2014 and 2018. 13 The protracted nature of the crisis and the dismal prospects for positive change have led to a considerable degree of critical reflection across the nexus from different perspectives and actors in WBGS. The UN notes that “humanitarian action extends to less traditional areas of intervention and calls for a much closer collaboration between humanitarian actors and the government” (UNSCO, 2016: 17). Along the same lines, the Humanitarian Response Plan for 2018 (OCHA, 2017a: 7 and 30) recognizes that “key drivers of vulnerability are common to both the humanitarian and development needs”. As noted by the Mapping and Synthesis of Evaluations carried 33The impact of assistance on poverty and food security in a fragile and protracted-crisis context groups have been identified as beneficiaries of both humanitarian and development inter- ventions, both of which must occur simultaneously in order to be effective. Humanitarian and development programming are increasingly aligned in order to provide durable and sustainable assistance capable of building resilience and reducing vulnerability. In other words, a blend of interventions tends in practice to prevail on a strict divide between humanitarian and development interventions, leveraging on the “humanitarian-develop- ment nexus” and operationalizing the so-called “new way of working” (OCHA, 2017b) as outlined in the UN Secretary-General’s Report for the World Humanitarian Summit (UN, 2016). The most important modalities of assistance in WBGS are: (i) in-kind provision of basic foodstuffs through baskets generally including wheat flour, rice, pulses and vegetable oil; (ii) food vouchers for use on selected items with designated merchants; and (iii) cash transfers distributed mostly through e-cards for cash disbursements. The aforementioned forms of assistance are listed in increasing flexibility, meaning that the mode of assistance provides a greater range of choice to targeted households, has cheaper implementation, and is less likely to focus on basic needs. Vocational training programs and other forms of livelihoods support can also help families rise above the poverty line. Other forms of support such as health and housing assistance are also quite important, especially in acute crisis (e.g. the 2014 war in Gaza). Assistance in Palestine is delivered by many actors. In terms of financial volume, major implementing actors include the Ministry of Social Development (MoSD), the United Nations Relief and Works Agency for Palestine Refugees in the Near East (UNR- WA), and the World Food Programme (WFP). While a large number of donors sup- port UNRWA’s activities, the two largest donors to direct assistance are the EU and, until 2017, the USA. Charities linked to zakat — the payment made under Islamic law on cer- tain kinds of property used for charitable and religious purposes — as well as assistance through de-facto authorities in the Gaza Strip are equally important sources of financial inflows (Culbert, 2017). While modalities of assistance vary between implementing bodies and beneficiary groups, selection criteria and program objectives are similar. The principal beneficiary selection tools used by actors for food and social assistance are poverty-based, using varia- tions of a proxy means testing formula. Institutional structures and political considerations are primary determinants in how social security assistance, direct food assistance and cash assistance are defined and channeled. Some development donors fund through governmental channels, such as the MoSD, while some humanitarian donors fund through humanitarian actors, such as UNRWA or WFP. As a result, the current system of delivering assistance is fragmented despite recent efforts working towards effective coordination between humanitarian and development actors (Culbert, 2017). The recent MoSD’s strategy (MoSD, 2017) holds promise in both coordinating and aligning assistance efforts of multiple actors by address- ing underlying social-economic challenges. However, this strategy remains at an early pol- icy stage. out by UNEG (2018: 28), in the occupied Palestinian territories there is recognition that “the scope of program- ming needs to transcend standard ‘good practice’ in order to mitigate the negative effects of what is likely to be a deteriorating situation”. 34 D. Romano et alii 3.3 Assistance to Palestinians in 2013-2014 Assistance to the WBGS is composed of a very heterogeneous set of modalities, implementing bodies and beneficiary groups, reflecting different conditions at the local level as well as between the West Bank and the Gaza Strip. Types of Assistance According to SEFSec (FSS-PCBS, 2016), approximately 40% of all Palestinian house- holds reported receiving at least one type of assistance in 2014. There is a marked dif- ference in the share of households receiving assistance in Gaza Strip (84%) compared to the West Bank (less than 17%) (Table1). Between 2013 and 2014, the share of assisted households in the Gaza Strip increased by more than 18%, even greater than the amount observed in 2011 (FAO-UNRWA-WFP, 2013). However, the increase in share of assisted households between 2013 and 2014 in the West Bank was less than 2%, standing 8 per- centage points below the level existing in the region in 2011. Table1 illustrates the prevalence of in-kind food, cash transfers and food vouchers provided to Palestinian households. Between 2013 and 2014, the composition of the vari- ous types of assistance in the West Bank did not change significantly, while composition of assistance in the Gaza Strip underwent important changes. In the West Bank, a large share of households reported that “Cash” and “In-kind food” were the two types of the assistance they received the most of in 2013 and 2014. On the other hand, the major cat- Table 1. Share of households receiving assistance by type of assistance and region, 2013-2014. WBGS West Bank Gaza Strip 2013 2014 2013 2014 2013 2014 In-kind food 24.6% 28.0% 7.5% 7.6% 57.5% 67.0% Health care 0.4% 2.3% 0.6% 2.7% 0.2% 1.6% Clothing 0.7% 2.1% 0.4% 0.3% 1.3% 5.7% Job creation 1.3% 0.3% 0.3% 0.2% 3.2% 0.6% Compensation martyrs 0.2% 0.3% 0.1% 0.3% 0.4% 0.4% Cash 16.8% 16.2% 10.5% 8.3% 28.9% 31.2% Health insurance 11.5% 7.8% 0.7% 1.2% 32.2% 20.3% Food vouchers 3.0% 8.2% 2.0% 1.6% 4.7% 20.8% School feeding 0.1% 0.1% 0.1% 0.0% 0.1% 0.4% Productive inputs 0.1% 0.0% 0.2% 0.0% 0.0% 0.1% Drinking water 0.4% 1.8% 0.0% 0.1% 1.0% 5.2% Electricity 0.2% 0.2% 0.2% 0.3% 0.0% 0.2% Housinga - 9.2% - 0.9% - 25.0% Other 0.6% 1.2% 0.2% 0.0% 1.3% 3.4% At least one form of assistance 32.4% 39.7% 15.2% 16.5% 65.7% 84.2% a Not included in the 2013 SEFSec survey. Source: FSS-PCBS (2016): Table 7.1, modified. 35The impact of assistance on poverty and food security in a fragile and protracted-crisis context egories of assistance reported in the Gaza Strip fluctuated between the two years. New types of assistance outside the three core types (“In-kind food”, “Cash” and “Health insur- ance”) were reported in the Gaza Strip. These included “Housing” (shelter, rent, caravan), “Food voucher”, “Drinking water” and “Clothing”. All four increased significantly between 2013 and 2014 in response to worsening living conditions as a result of the armed conflict. Value of Assistance In 2014, assisted households received an average of 102 US$/month. However, nation- al averages mask significant regional differences in both levels and trends. Table 2 reports the average monthly value received by households in the two regions for each types of assistance during 2012-2014. There was a general decline in the average value of assistance for cash and food in the West Bank from 2013 to 2014. Conversely, assistance for employ- ment and provision of agricultural inputs increased. In the Gaza Strip the average value of support increased for many assistance types but food assistance that did not change much. Employment assistance represented the largest average allowances given to households in 2014. Among “Other” forms of support, the largest average values are seen for housing and shelter assistance. Support to agricultural production activities almost disappeared in Gaza Strip after 2012. The value of assistance varies across different types of households (Table 3). Support to refugee households was slightly greater than that of non-refugee households (107 vs. 91 US$/month). Moreover, a substantial difference was recorded in 2014 based on gen- der household heading: female-headed households received on average 30% more support than male-headed households (127 vs. 98 US$/month). This reveals that female-headed households are more dependent on assistance, probably due to higher vulnerability. The composition of assistance across different household typologies emphasizes the different needs of various groups (Table 3). Female-headed households are more likely to receive assistance in the form of cash and free health services than male-headed house- Table 2. Average value of support by type of assistance, US$/month. Type of assistance West Bank Gaza Strip 2012 2013 2014 2012 2013 2014 Cash 115 79 55 95 92 123 In-kind food 45 34 27 37 36 48 Food vouchers 42 43 28 30 48 32 Job creation 115 97 126 82 147 215 Agricultural inputs 46 69 123 129 na 9 Housing na na 231 na na 211 Othera 71 70 135 4 17 110 Average per assisted household 128 96 86 65 102 108 a The “Other” category in years 2012 and 2013 includes also housing. Source: FSS-PCBS (2016): Table 7.3, modified. 36 D. Romano et alii holds. This is probably due to the demographic composition of the former, with a major- ity of households headed by widows and elderly women. The comparison between refu- gee and non-refugee indicates a cash preference by non-refugee households, while refugee households receive a larger share of assistance in “Other” forms, including substantial sup- port for housing. Sources of Assistance Social assistance coverage increased between 2013 and 2014, reflecting deteriorating livelihood conditions–especially in the Gaza Strip, where more than four households out of five were receiving assistance in 2014. Overall, reported sources of assistance are given primarily by the Palestinian Ministry of Social Affairs (currently renamed the Ministry of Social Development, or MoSD), UNRWA, international agencies, charitable and religious associations, and informal assistance (family, relatives or friends). However, key differenc- es are observed between the West Bank and the Gaza Strip (Table 4). In the West Bank, 7% of households reported receiving assistance from the Ministry of Social Affairs in 2014, a slightly lower figure than that reported in 2013 (8%). The other two most cited sources of assistance in 2014 were UNRWA and informal assistance (fam- ily and relatives), which remained unchanged from 2013 levels. A different picture emerges from the data in the Gaza Strip. Not surprisingly, the larg- est source of social assistance in 2014 was UNRWA, an organization that provided food assistance to some 867,000 refugees. A number of other sources of assistance were report- ed, including the Palestinian Ministry of Social Affairs, international agencies, charitable and religious associations, worker unions, and family and friends. Informal sources of social assistance more than halved, dropping to 7% in 2014. This is a clear sign that infor- mal social networks were unable to help in times of widespread severe hardship caused by the war. Table 3. Composition of assistance by region and household group, share of total value received, 2014. Type of support West Bank Gaza Strip Refugee Non- refugee Male- headed Female- headed Cash 36.4% 34.5% 31.8% 40.2% 34.0% 40.4% In-kind food 15.3% 26.8% 23.6% 24.7% 25.7% 15.6% Health insurance 19.8% 0.8% 5.3% 5.0% 3.1% 16.2% Food vouchers 3.1% 5.5% 4.7% 5.7% 5.5% 2.3% Housing 13.1% 21.6% 24.4% 12.2% 20.9% 12.9% Other 0.1% 5.6% 5.5% 2.2% 5.0% 0.7% Remaining sources 12.2% 5.2% 4.7% 10.0% 5.8% 11.8% Average per assisted household (US$/month) 86 108 107 91 98 127 Source: FSS-PCBS (2016): Table 7.4, modified. 37The impact of assistance on poverty and food security in a fragile and protracted-crisis context 4. Poverty and Food Security The profiling of Palestinian households in terms of poverty quartiles before receiv- ing assistance shows expected patterns14 (Table 5): moving from poorer to richer house- holds saw a parallel decrease in household size, an increase in educational attainment, a decrease in the dependency ratio, and an increase in the employment rate (including that of the head of the household). Poverty in the WBGS is determined by the employability of household members. Food security on the other hand is largely influenced by access dimension, specifically by individuals’ labor entitlement. Table 6 therefore provides a detailed account of household heads’ labor indicators across poverty quartiles. By and large, poorer households had more problematic labor conditions. For instance, household heads who worked fewer hours were more likely to be poor, just as irregular employment and lower level occupations were more related to poverty. Usually, poverty is correlated to employment in the primary and construction sectors. In short, heads of poorer households tend to have more infor- mal and irregular jobs that do not require high levels of formal skills and/or education, such as jobs in basic production sectors. 14 Only the female-headed household share does not show a clear pattern. Another characteristic (not reported in the table) that does not change at all is the number of sources of income per household: on average, two per household. Table 4. Reported sources of assistance by Regiona. West Bank Gaza Strip 2013 2014 2013 2014 Ministry of Social Affairs 8.2% 6.8% 19.6% 23.5% Other PA agencies 0.9% 2.0% 4.2% 8.6% Political parties 0.0% 0.1% 0.4% 8.6% Zakat/other religious institutions 0.5% 0.6% 0.5% 2.7% International agencies (excluding UNRWA) 1.4% 1.2% 9.3% 21.3% UNRWA 2.1% 4.0% 42.6% 62.3% Arab countries 0.0% 0.1% 0.3% 2.8% Charity/religious 0.4% 0.3% 3.8% 19.5% Family and relatives 2.8% 2.8% 14.8% 6.8% Friends/Neighbors 1.1% 0.9% 1.8% 4.8% Workers union 0.0% 0.0% 21.6% 12.9% National banks 0.0% 0.0% 0.0% 0.5% Local reform commission 0.0% 0.0% 0.1% 0.6% Other 0.4% 0.9% 0.3% 3.3% Any type of assistance 15.2% 16.5% 65.7% 84.2% a Sources of assistance are not mutually exclusive. Some households reported receiving assistance from more than one source. Source: FSS-PCBS (2016): Table 7.5. 38 D. Romano et alii As expected, there is a direct relationship between poverty and food insecurity (Table 7). This is measured by the Food Consumption Score (FCS) and the Household Food Insecurity Access Scale (HFIAS), two proxies for the qualitative and quantitative dimen- Table 5. Households’ profile per poverty quartile, 2013. Q1 Q2 Q3 Q4 Total Average household size 7.7 5.2 4.8 4.5 5.6 Share of HH with female head 6.4% 11.5% 11.4% 9.1% 9.6% Share of HH with head with secondary education or above 28.1% 34.2% 39.1% 51.0% 38.1% Global dependency ratio 1.20 1.19 1.02 0.90 1.08 Share of HH whose head does not work 28.9% 28.4% 23.5% 22.4% 25.8% Household employment rate 32.1% 36.9% 40.5% 43.7% 38.3% Authors’ elaboration on SEFSec 2014 data. Table 6. Head of household employment statistics per poverty quartile, 2013. Q1 Q2 Q3 Q4 Total Working Status           Employed from 1-14 hours 5.1% 4.6% 2.5% 1.3% 4.2% Employed 15-34 hours 6.1% 6.9% 5.1% 3.2% 6.0% Employed 35 hours and over 41.7% 46.5% 58.5% 63.5% 47.7% Temporarily absent 14.6% 10.6% 6.6% 3.9% 11.2% Looked for a job (already worked) 6.9% 3.9% 1.2% 2.1% 4.6% Looked for a job (never worked) 2.1% 2.6% 0.7% 1.1% 1.9% Did not look for work because of frustration 0.7% 0.9% 0.6% 0.6% 0.7% Full time student 0.1% 0.0% 0.0% 0.0% 0.0% Housewife 4.0% 5.0% 4.9% 3.4% 4.4% Unable to work 16.8% 14.7% 12.8% 8.1% 14.8% Other 0.0% 0.0% 0.2% 0.0% 0.0% Professional Status Employer 2.4% 2.2% 3.9% 11.9% 5.1% Self-employed 11.3% 11.5% 12.4% 13.9% 12.3% Unpaid family worker 0.2% 0.1% 0.3% 0.1% 0.2% Waged employee 61.9% 59.9% 60.5% 52.2% 58.6% Sector of employment Agriculture, fishing and forestry 8.7% 6.7% 3.5% 2.2% 5.3% Mining, quarrying and manufacturing 6.5% 8.6% 11.5% 13.6% 10.0% Construction 18.2% 16.7% 16.3% 12.6% 16.0% Commerce, restaurants and hotels 10.9% 11.2% 14.6% 20.2% 14.2% Transportation, storage and communication 7.2% 5.5% 6.1% 5.1% 6.0% Services and other activities 24.3% 24.9% 25.1% 24.5% 24.7% Authors’ elaboration on SEFSec 2014 data. 39The impact of assistance on poverty and food security in a fragile and protracted-crisis context sions of food security, respectively (cf. section 5.1). Probably the most striking indicator related to poverty is the share of households receiving assistance. This value encompasses almost two thirds of all households in the lowest quartile and 7.6% of households in the highest quartile. Both indicators of food security show the expected regularities in that poorer households have lower FCS values. Meanwhile, poorer households have larger shares of poor or borderline FCS (Q1 three times larger than that of Q4) as well as insuf- ficient dietary quantities (HFIAS in Q1 eight times larger than that of Q4). Quite surpris- ingly, the average value of assistance rapidly decreases from the lowest to the second-low- est quartile, but then increases again in the two higher quartiles15. 5. Data and methods 5.1 Data The Socio-Economic and Food Security (SEFSec) survey has been administered since 2009 to monitor the status of food security among Palestinian households. The SEFSec methodology accounts for the multi-dimensional drivers of food insecurity in WBGS by exploring topics such as asset-based poverty, food consumption, and resilience. This is done to capture the capacity households have to adapt, transform and cope with shocks. Besides these three main pillars, the questionnaire collects data on aspects such as socio-demo- graphics, assistance, expenditure and consumption, all of which are useful for the analysis. The dataset includes data from the fifth and sixth SEFSec surveys. Data collection took place in 2014 and 2015, with a reference period covering the six months preceding the interview (the second half of 2013 and 2014, respectively). The 2013 SEFSec survey was conducted on a sample of 7,503 households (4,949 in the West Bank and 2,554 in the Gaza Strip), while the 2014 sample included 8,177 households (5,047 in the West Bank and 3,130 in the Gaza Strip). The samples are representative for various levels of disag- gregation, including gender, refugee status, governorate, locality type (i.e. urban, rural and refugee camp) and, for the West Bank only, Areas A/B and C (FSS-PCBS, 2016). An important feature of the 2013-2014 SEFSec is that 92% of the households inter- viewed in 2013 were included also in the 2014 wave. Therefore, a sample of 6,881 units 15 However, this seems to be related to the higher average value of assistance in the West Bank to households that own some type of business: essentially, it is a support to investment that is able to generate employment. Table 7. Households’ assistance and food security status per poverty quartile, 2013. Q1 Q2 Q3 Q4 Total Per capita expenditure (NIS/month) 305 461 593 860 554 Share of HH receiving assistance 62.5% 41.8% 21.3% 7.6% 33.3% Average value of assistance per HH (NIS/month) 418 293 347 321 368 Households with insufficient dietary quantity (HFIAS) 50.7% 29.3% 14.8% 6.4% 25.3% Households with poor or borderline FCS 30.4% 26.4% 17.8% 10.2% 21.2% Average household FCS 70 72 76 80 74 Authors’ elaboration on SEFSec 2014 data. 40 D. Romano et alii (4,454 in the West Bank and 2,427 in the Gaza Strip) can be used to analyze the impact of assistance on Palestinian households through the panel structure of the dataset. The main variables used in the analysis are summarized in Table 8. They include the three outcome variables of interest: a measure of poverty and two measures of food secu- rity (i.e. HFIAS and FCS, the latter also broken down in its main components), a set of household socio-demographics that are the usual correlates used to analyze the outcomes, and some geographical dummies to account for regional/residence differences used to capture any unobserved heterogeneity.16 Poverty outcomes are measured as an asset-based poverty index closely related to living standards. An asset-based poverty index better reflects long-term wealth over an expenditure-based poverty index, a short-term measure which in principle would work better in an impact evaluation of aid effectiveness. Additionally, the asset-based pover- ty index was chosen since total household expenditure is not accurately sampled by the SEFSec questionnaire. Indeed, an assessment commissioned by SEFSec administrators to evaluate the robustness and reliability of expenditure-based poverty measures resulted in the decision to abandon money-based (i.e. expenditure) measures of poverty because they were inconsistent with similar measures based on benchmark data from the Palestine Expenditure and Consumption Survey of 2011 (PECS) (Langworthy et al., 2014; Smith, 2014).17 Furthermore, in the context of protracted crisis such as the currently ongoing one in Palestine, assistance becomes a key source of income for the majority of households, establishing itself as a “structural” component of household income. Assistance has sig- nificantly contributed to building household assets over the years and helps maintain a given level of standards of living via consumption smoothing. If assistance to households decreases, household assets would decrease in response because the household sells its assets to countervail the reduction in assistance. Food security is proxied by two measures, namely the Household Food Insecuri- ty Access Scale (HFIAS), a quantitative measure of the dimension of food consumption (Coates et al., 2007), and the Food Consumption Score (FCS) that captures the quality of household diets (WFP, 2008). HFIAS is an indicator based on responses to nine ques- tions, five of which relate to the size and frequency of meals consumed in the 30 days preceding the survey. HFIAS is value ranging from 0 to 27, where a higher score indicates an insufficient dietary quantity. FCS is an indicator based on the number of days specific food groups are consumed in the seven days preceding the survey. The FCS is a continu- ous score where a value less than or equal to 45 or between 45 and 62 respectively indi- cate poor or borderline food consumption. This value is obtained by assigning a specific weight to each food group in accordance to its contribution to dietary quality. 16 The variables listed in Table 3.1 are the ones actually used in the following analysis, that is they are only a subset of the wider set of candidate variables that in principle could be used. Unfortunately, the SEFSec survey is designed only to monitor the evolution of food security in Palestine. As such it does neither have the wealth of variables that can be usually found in a standard multi-purpose survey (e.g. household cultural traits, house- hold behavior other than food consumption, etc.), nor the depth of data typical of household expenditure/con- sumption surveys (e.g. detailed information on household expenditures, food consumption composition, etc.). 17 The overall conclusion of these studies was that “in the absence of other options, an asset-based measure of poverty can thus serve as a valid, stand-alone measure for the purposes of the SEFSec food insecurity analysis.” (Smith, 2014: 21). 41The impact of assistance on poverty and food security in a fragile and protracted-crisis context The pros and cons of these two indicators have been assessed in several review and validation studies of food security indicators (Carletto et al., 2013). IFPRI (2006) con- cluded that the FCS weighting system for the food frequency scores might not be able to accommodate variations across space and time. Nevertheless, IFPRI found positive associ- ations between FCS values and caloric consumption per capita in some studies. The infor- mation generated by HFIAS is used to assess the prevalence of household food security and detect changes over time. Moreover, validations conducted in Latin America and sub- Saharan Africa (Melgar-Quinonez et al., 2006; Knueppel et al., 2010) found that the indi- cator demonstrated reliability and validity in the local contexts in which it was deployed. Table 8. Summary statistics of key variables. Variable Meaning Mean Standard deviation Min max l_ass_index Log of asset based poverty index 7.09 0.33 5.52 8.28 fcs Food consumption score (FCS) 74.28 17.06 0.00 112.00 hfias Household Food Insecurity Access Scale (HFIAS) score 4.64 6.56 0.00 27.00 vegfru_fcs FCS cereals, tubers, pulses, vegetable and fruit 26.96 4.93 0.00 49.00 meatmilk_fcs FCS meat and milk 40.85 14.65 0.00 56.00 oilsug_fcs FCS fats and sugar 6.46 1.13 0.00 7.00 mass log of HH monthly assistance 1.96 2.63 0.00 10.82 ydum dummy for year 2014 0.50 0.50 0.00 1.00 massy interaction mass*ydum 1.04 2.12 0.00 10.82 lhsize Log of household size 1.81 0.42 0.69 3.30 lexp Log of household monthly expenditure (NIS) 7.72 0.75 1.79 11.16 dep_ratio Dependency ratio (aged 0-15+aged >65)/aged 15-65 1.10 1.34 0.00 7.00 rat_emp % of employed people aged >15 in the HH 0.37 0.24 0.00 1.00 agehead Age of HH head (years) 45.34 14.37 19.00 98.00 femhead HH head gender (female = 1) 9.66%   0 1 head_ref HH head status (refugee = 1) 41%   0 1 high_ed HH head education (secondary education or higher = 1) 38.12%   0 1 employed HH head occupational status (employed = 1) 70.42%   0 1 qly_deprived HH with low FCS (< 61) (yes = 1) 22.26%   0 1 qty_deprived HH with insufficient food intake, HFIAS (yes = 1) 23.21%   0 1 ass HH receiving assistance (yes = 1) 37.71%   0 1 WB North Regional dummy (West Bank North = 1) 27.58%   0 1 WB Center Regional dummy (West Bank Center = 1) 17.69%   0 1 WB South Regional dummy (West Bank South = 1) 19.46%   0 1 GS North Regional dummy (Gaza Strip North = 1) 18.47%   0 1 GS Center Regional dummy (Gaza Strip Center = 1) 5.19%   0 1 GS South Regional dummy (Gaza Strip South = 1) 11.61%   0 1 rural Locality of residence (rural = 1) 18.62%   0 1 camp Locality of residence (refugee camp = 1) 9.74%   0 1 urban Locality of residence (urban = 1) 71.64%   0 1 42 D. Romano et alii Besides the considerations above, the SEFSec dataset does not include enough data to build other food security indicators such as the food caloric intake. 5.2 Methods To estimate the impact of assistance on a given dimension of well-being, such as pov- erty or food security, we need to control for possible unobserved heterogeneity in partici- pation in the assistance program. Due to the targeting strategies of the different agencies that provide assistance to Palestinian households, treated households are quite different from untreated ones. Notably, the probability of receiving assistance is correlated with a set of characteristics mostly related to poverty (cf. section 4). As a result, the selection bias is likely to be pervasive (Khandker et al., 2010). Moreover, further unobserved targeting variables may affect both the outcome variable and the probability to receive assistance. Building on the panel structure of SEFSec dataset, we used a difference-in-difference (DD) approach to get rid of aforementioned biases. The DD model assumes that the het- erogeneity in participation is fundamentally time invariant once conditioned on a set of household characteristics (X): E(Y0t – Y0t-1 | T = 1,X) = E(Y0t – Y0t-1 | T = 0,X) (1) where Y0t is the potential outcome without the treatment measured at time t. T is the treatment status, which equals to 1 if the household received assistance and 0 otherwise. The assumption of time invariant heterogeneity implies that the dynamics observed in the control group are the same as the ones observed in the treated group had the latter not been treated. Unfortunately, the SEFSec dataset does not allow testing for the “parallel trend” hypothesis. However, considering the short time distance between the two SEFSec waves, the risk that this assumption does not hold is low. In regression form the DD estimator is given by: Yi,t = αi + βTi + γt + δTit + ∑ζXi,t + εi,t (2) where t is a time dummy (1 in the second period, 0 otherwise). Ti is the treatment dum- my, with a value of 1 for the treatment group and 0 for the control. The casual effect of the treatment is assumed to be additive. In the classical DD model, the δ parameter — which is associated with the interaction term between the treatment Ti and the time dummy var- iable t — identifies the expected impact (Angrist and Pischke, 2008). The traditional DD regression uses dichotomic (i.e. treated/non-treated) treatment variables. However, continuous treatment variables measuring the intensity of the treat- ment can be also used (Card, 1992; Acemoglu et al., 2004). Continuous variables fully exploit the information content of available data. For the purpose of this study, the most suitable candidate is the monthly value of assistance received by the household. In this case, it can be demonstrated that for the i-th household the δ parameter is equivalent to: δ = (Yi1 – Yi0 | Ti = Ti1,Xi) – (Yi1 – Yi0 | Ti = Ti0,Xi) (3) 43The impact of assistance on poverty and food security in a fragile and protracted-crisis context where the numerator is the difference in outcome variation over time given the final and initial values of the continuous intervention variable and the denominator is the difference between the final and the initial value of the continuous treatment variable. In the case of an increase of the continuous treatment variable between the two periods, a positive value of δ indicates that the increased treatment intensity determines a higher increase of the outcome variable. This implies that the impact of the treatment is positive. Moreover, thanks to the time dimension of the panel, we can include in (2) household specific intercepts or fixed effect, αi. Irrespective of the adopted fixed effect estimator, this is equivalent to including a dummy variable for each household in equation (2) (Wool- dridge, 2013). Equation (3) will still hold provided that we condition on both X and αi. The key identifying assumption in this context is that treatment intensity is not cor- related with individual unobserved trends, although it can correlate with individual per- manent characteristics. We posit that the intensity of assistance (“mass”, measured in log- arithms) impacts the outcome variable, i.e. either the log of poverty asset index (“l_ass- index”) or one of the food security indicators (“hfias” or “fcs”). The intensity of assistance and the outcome variable are both affected by a set of household characteristics that we assume to be time-invariant, including location, refugee status, and education of the head of household. All of these are captured by αi. We further conditioned on potential time variant confounders such as dependency ratio, household size, ratio of employed house- hold members to the number of household members of working age, and employment status of the head of the household. In the case of poverty models potential endogeneity may remain even after having conditioned on the fixed effects due to the nature of the targeting process. Therefore, we implemented the 2SLS version of both the pooled OLS and the fixed effect estimators. In the case of food security indicators, we can assume that regressors are exogenous because targeting is made on poverty, not on food security indi- cators. Noticeably, in the case of the HFIAS score, we have to deal with a censored variable whose distribution has a clear peak at zero. In such a case the fixed effects tobit model estimates would be affected by the so-called “incidental parameters” problem especially in case of short time panel datasets (Greene, 2004). To ensure consistency with the fixed effect models of continuous outcome variables (asset-based poverty index and FCS), in the case of HFIAS model we used the semi-parametric estimator of fixed effect tobit mod- els proposed by Honoré (1992), which is consistent and asymptotically normal even for time dimension of 2 as in our case. 6. Results We first run a pooled OLS regression using a sandwich estimator of the covariance matrix. Results in the case of the asset-based poverty index18 are reported in the first two columns of Table 9. All independent variable parameters except for a few regional dum- mies are significant at p=0.05. Both the household size and the dependency ratio affect the index negatively, while the ratio of employed household members over working age 18 The dependent variable – i.e., the log of the asset-based poverty index – is built in such a way a higher index value corresponds to wealthier households. This should be considered when interpreting the results in Table 9. 44 D. Romano et alii household members shows a clear positive effect. This confirms that poverty is mostly a matter of (a lack of ) employability. The characteristics of head of households that positive- ly impact the index are the following: education, age, employment status, refugee status or living in the West Bank. On the other hand, households situated in rural areas and refugee camps negatively impact the outcome variable. All estimates have expected signs: higher educational attainment, employment and living in the West Bank over the Gaza Strip all decrease the chances that a household is poor. Conversely, holding refugee status or living far away from an urban center increases the likelihood of being poor. The impact denoting the intensity of assistance is captured by the interaction term “massy”. The value of monthly assistance positively impacts the asset-based poverty index. Table 9. Asset-based poverty index regression models, Palestine. Pooled OLS Pooled 2SLS Fixed Effect Fixed Effect IV Coef. Student’s t Coef. z Coef. Student’s t Coef. z massa -0.03 -24.96 -0.03 -25.33 -0.02 -15.56 -0.02 -15.07 ydum -0.01 -1.89 -0.01 -1.91 -0.03 -5.18 -0.03 -5.07 massya 0.00 2.06 0.00 2.11 0.01 6.04 0.01 5.75 lhsize -0.31 -54.19 -0.31 -54.22 -0.27 -37.46 -0.27 -37.47 dep_ratio -0.02 -14.26 -0.02 -14.23 -0.02 -10.77 -0.02 -10.77 rat_emp 0.10 8.53 0.10 8.47 0.15 12.29 0.15 12.32 employed 0.04 5.55 0.03 5.35 -0.01 -4.20 -0.01 -4.20 agehead 0.00 10.45 0.00 10.42       refhead 0.04 8.5 0.04 8.47         femhead -0.02 -2.77 -0.02 -2.70         high_ed 0.09 19.05 0.09 18.85         WB North 0.13 13.33 0.12 12.84         WB Center 0.23 22.58 0.23 21.98         WB South 0.09 9.25 0.09 8.83         GS North -0.01 -1.01 -0.01 -1.05         GS Centerb                 GS South -0.01 -1.03 -0.01 -1.01         rural -0.10 -16.06 -0.10 -15.98         camp -0.04 -5.06 -0.03 -5.01         constant 7.44 417 7.45 415.01         R2 0.45       0.36       KP rk under-identification ChiSq p=0.00 CD Wald F >350 >350 HJ over-identification ChiSq exactly id. exactly id. IV (excluded) ass, assy   ass, assy   F test of fixed effect         1.8 p=0.00     a This variable has been instrumented; b GS Center, where Gaza City is located, is assumed as reference. Note: KP is the Kleibergen-Paap LM test for under-identification of the model; CD is the Cragg Donald weak identification test; HJ is the Hansen J statistics for over-identification of the model (cf. Baum et al., 2007). 45The impact of assistance on poverty and food security in a fragile and protracted-crisis context However, despite being statistically significant, the coefficient estimate is close to 0. To deal with possible endogeneity, we performed a pooled 2SLS instrumenting the variable and the interaction term with dummies for assistance and its interaction with time. How- ever, the size of the coefficient of the interaction term does not change in the case of 2SLS. In order to account for unobserved individual heterogeneity, we run a fixed effect regression. This is done because the Hausmann test rejected the hypothesis of absence of correlation between random effects and regressors. Table 9 reports the parameter esti- mates obtained with the fixed effect estimator on transformed data as deviations from the group means.19 We also implemented the corresponding 2SLS version for the fixed effect estimator using the same instruments employed in the pooled model (last two columns of Table 9). All time-invariant regressors are perfectly correlated with the household specific intercepts, therefore only the time varying variables are considered in the fixed effect models: dependency ratio, household size, ratio of employed household members to working age members, and employment status of household head. Both models con- firm that the intensity of assistance has a significant effect in reducing household poverty. In all the models, the coefficients of the interaction term are statistically significant stable around 0.01: a 10% increase of assistance on average leads to a direct 0.1% increase of the asset-based index. To take into account the fact that the West Bank and the Gaza Strip are physically, politically and economically apart, we estimated the impact of assistance separately for the two regions (Table 10 and 11, respectively). As expected, the impact is significantly posi- tive in the West Bank and of the same order of magnitude as Palestine as a whole (Table 9). This was true after having accounted for individual heterogeneity. Quite surprisingly, we obtained a non-significant impact of assistance in the Gaza Strip. This seems related to the very peculiar situation present in Gaza. In 2014, more than four households out of five received assistance (cf. section 3.3), largely irrespective of the household characteristics.20 This was done in order to offset the region’s widespread humanitarian crisis resulting from a ten-year long blockade and generalized “de-devel- opment” (UNCTAD, 2017). To make matters worse, a series of military operations took place over the last decade, ultimately culminating in the devastating war of July-August 2014 — exactly during the second period surveyed. This is likely to have blurred the caus- al relationship between assistance and poverty. The estimates in the case of HFIAS show the expected signs.21 In the models for Pal- estine as a whole (Table 12), the coefficient of the interaction term is significantly negative in the simple pooled OLS model as well as in models addressing the censored nature of the HFIAS variable. This means that assistance has a significant positive impact in ensur- 19 With this transformation we get rid of the large number of group dummies that would be included in the least square dummy variable estimator had the transformation not being made (Baltagi, 2005). 20 The poverty headcount ratio in the Gaza Strip is 53.0% while one third of population (33.8%) lives in extreme poverty according to monthly consumption patterns (PCBS, 2018a). According to Atamanov and Palaniswamy (2018) more than 90% of the bottom 40% in the Gaza Strip receive some form of aid; and even among the most well-off, half receive assistance. Another anecdotal evidence of the generalized humanitarian crisis is the higher concentration around the mean of average assistance per household in Gaza Strip vis-à-vis West Bank with the latter having a coefficient of variation that is five times larger than the former. 21 HFIAS is a measure of quantity deprivation of food showing higher scores the lesser the food consumed by the household. 46 D. Romano et alii ing the consumption of adequate quantities of food. Moreover, being a refugee, employed, well-educated, younger household head reduces household food insecurity. Regional models tell the same story, although it is worth noting that the impact of assistance is much stronger in the Gaza Strip than in the West Bank. This confirms the key role of assistance to ensure food security in a humanitarian crisis context such as the Gaza Strip, where two third of households receive in-kind food assistance and one fifth of sur- veyed households received food vouchers (cf. Table 1). In the West Bank, households have a wider portfolio of coping strategies available to them, including non-assistance strategies. Both regions have marked sub-regional differences. The governorates of the two main economic centers – Ramallah and East Jerusalem in the West Bank and Gaza City in the Table 10. Asset-based poverty index regression models, West Bank. Pooled OLS Pooled 2SLS Fixed Effect Fixed Effect IV Coef. Student’s t Coef. z Coef. Student’s t Coef. z massa -0.03 -16.29 -0.03 -17.15 -0.02 -10.01 -0.03 -10.15 ydum -0.02 -3.02 -0.02 -3.01 -0.03 -5.55 -0.03 -5.42 massya 0.00 0.53 0.00 0.47 0.01 3.61 0.01 3.34 lhsize -0.30 -38.49 -0.30 -38.49 -0.23 -25.1 -0.23 -25.13 dep_ratio -0.02 -10.99 -0.02 -10.94 -0.02 -8.25 -0.02 -8.24 rat_emp 0.09 5.91 0.09 5.85 0.16 9.82 0.15 9.72 employed 0.05 5.89 0.05 5.66 -0.02 -4.01 -0.02 -3.97 agehead 0.00 8.03 0.00 8.00       refhead 0.05 7.31 0.05 7.26         femhead -0.02 -1.86 -0.02 -1.77         high_ed 0.10 15.75 0.10 15.68         WB North -0.11 -14.93 -0.11 -14.85         WB Centerb                 WB South -0.15 -18.74 -0.15 -18.64         rural -0.11 -16.37 -0.11 -16.25         camp -0.07 -5.69 -0.07 -5.63         constant 7.63 345.45 7.63 345.47         R2 0.31       0.21       KP rk under-ident. ChiSq   1083 p=0.00 1013 CD Wald F >350 >350 HJ over-identific. ChiSq exactly id. exactly id. IV (excluded) ass, assy,   ass, assy   F test of fixed effect         1.8 p=0.00     a This variable has been instrumented; b WB Center, where Ramallah and East Jerusalem are located, is assumed as reference. Note: KP is the Kleibergen-Paap LM test for under-identification of the model; CD is the Cragg Donald weak identification test; HJ is the Hansen J statistics for over-identification of the model (cf. Baum et al., 2007). 47The impact of assistance on poverty and food security in a fragile and protracted-crisis context Gaza Strip – perform on average better than other districts. We do not have econometric evidence to explain this. However, we can argue that this happens for different reasons on the basis of secondary information. For instance, in the case of the West Bank, residing within the municipality of Ramallah or close to it is an advantage in terms of employment and market opportunities. Furthermore, the impact of Israeli settlements and territorial frag- mentation is less pronounced in these areas compared to WB North and WB South. For the Gaza Strip, residing close to the decision-making center of the de facto ruling authority and further away from the Israeli border22 is an advantage in terms of food security. 22 Israeli forces enforce a buffer zone by land and sea, the “access restricted areas”. According to Israeli authori- ties, up to 100 meters from the double wired/concrete fence built along the Gaza-Israel border is a “no go” area and up to 200 meters there is no access for heavy machinery. However, “humanitarian partners in the field have Table 11. Asset-based poverty index regression models, Gaza Strip. Pooled OLS Pooled 2SLS Fixed Effect Fixed Effect IV Coef. Student’s t Coef. z Coef. Student’s t Coef. z massa -0.03 -14.98 -0.03 -14.38 -0.02 -9.32 -0.02 -8.74 ydum 0.03 2.67 0.03 2.17 0.00 -0.11 0.00 -0.16 massya 0.00 -1.5 0.00 -1.08 0.00 1.34 0.00 1.15 lhsize -0.35 -43.49 -0.35 -43.28 -0.33 -33.63 -0.33 -33.69 dep_ratio -0.02 -9.1 -0.02 -9.11 -0.02 -6.81 -0.02 -6.82 rat_emp 0.14 7.86 0.14 7.86 0.12 7.46 0.12 7.49 employed 0.00 -0.17 0.00 -0.20 -0.01 -1.45 -0.01 -1.46 agehead 0.00 7.82 0.00 7.80       refhead 0.02 3.06 0.02 3.06         femhead -0.02 -2.13 -0.02 -2.11         high_ed 0.07 11.34 0.07 11.14         GS North 0.00 0.36 -0.01 -1.05         GS Centerb                 GS South 0.00 -0.09 0.00 -0.08         rural 0.01 0.4 0.01 0.40         camp -0.01 -0.9 -0.01 -0.88         constant 7.52 314.63 7.52 312.85         R2 0.48       0.46       KP rk under-identific. ChiSq p=0.00 CD Wald F >350 >350 HJ over-identification ChiSq exactly id. exactly id. IV (excluded) ass, assy   ass, assy   F test of fixed effect         1.5 p=0.00     a This variable has been instrumented; b GS Center, where Gaza City is located, is assumed as reference. Note: KP is the Kleibergen-Paap LM test for under-identification of the model; CD is the Cragg Donald weak identification test; HJ is the Hansen J statistics for over-identification of the model (cf. Baum et al., 2007). 48 D. Romano et alii Ta b le 1 2. H FI A S re g re ss io n m o d el s. V ar ia bl es Pa le st in e W es t B an k G az a St ri p Po ol ed O LS To bi t H on or é es tim at or Po ol ed O LS To bi t H on or é es tim at or Po ol ed O LS To bi t H on or é es tim at or C oe f. St ud en t’s t C oe f. z C oe f. z C oe f. St ud en t’s t C oe f. z C oe f. z C oe f. St ud en t’s t C oe f. z C oe f. z m as s 0. 82 22 .8 2 1. 17 22 .3 7 0. 93 12 .6 2 0. 61 12 .0 4 0. 61 12 .0 4 0. 97 7. 27 0. 97 16 .4 6 1. 40 16 .8 0 1. 18 11 .8 4 yd um -1 .3 4 -1 3. 69 -2 .5 1 -1 1. 08 -3 .4 5 -1 2. 14 -1 .4 7 -1 4. 93 -1 .4 7 -1 4. 93 -4 .4 5 -1 3. 29 -0 .6 9 -1 .9 9 0. 24 0. 43 0. 90 1. 33 m as sy -0 .4 0 -9 .7 3 -0 .2 6 -4 .2 2 -0 .3 5 -4 .5 3 -0 .2 1 -3 .2 4 -0 .2 1 -3 .2 4 -0 .3 5 -2 .1 -0 .5 8 -7 .5 1 -0 .7 8 -6 .8 4 -1 .1 6 -8 .1 lh si ze 2. 43 19 .8 2 4. 51 19 .7 1 3. 90 11 .1 7 1. 73 13 .1 3 1. 73 13 .1 3 4. 05 7. 42 3. 62 14 .1 9 4. 71 13 .7 5 3. 95 8. 38 de p_ ra tio 0. 05 1. 36 0. 15 2. 27 0. 10 0. 99 0. 02 0. 46 0. 02 0. 46 0. 11 0. 7 0. 16 1. 85 0. 25 2. 16 0. 10 0. 67 ra t_ em p -1 .8 4 -8 .0 8 -3 .6 8 -7 .7 7 -3 .3 7 -4 .7 -1 .0 4 -4 .2 7 -1 .0 4 -4 .2 7 -3 .3 6 -3 .2 6 -3 .7 9 -7 .6 3 -5 .5 9 -7 .7 5 -3 .4 5 -3 .2 9 em pl oy ed -0 .9 3 -6 .5 1 -1 .3 0 -5 .1 5 -1 .3 2 -3 .3 6 -0 .8 5 -5 .4 4 -0 .8 5 -5 .4 4 -1 .5 2 -2 .4 8 -0 .6 8 -2 .3 9 -0 .4 1 -1 .1 3 -1 .1 7 -2 .2 1 ag eh ea d -0 .0 3 -9 .1 3 -0 .0 5 -8 .1 9 -0 .0 2 -5 .0 3 -0 .0 2 -5 .0 3 -0 .0 6 -8 .1 6 -0 .0 8 -7 .7 0 re fh ea d -0 .2 5 -2 .2 6 -0 .8 1 -3 .9 4   -0 .4 1 -3 .4 1 -0 .4 1 -3 .4 1   -0 .2 1 -0 .9 4 -0 .2 7 -0 .9 2   fe m he ad 0. 18 0. 99 0. 67 2. 05   0. 36 1. 86 0. 36 1. 86   -0 .0 5 -0 .1 3 -0 .0 4 -0 .0 9   hi gh _e d -1 .3 2 -1 3. 69 -2 .6 6 -1 3. 89   -0 .8 7 -8 .8 8 -0 .8 7 -8 .8 8   -2 .0 2 -9 .4 4 -2 .6 8 -9 .4 4   W B N or th -1 .1 0 -4 .4 3 -2 .7 9 -7 .0 0   0. 71 6. 04 0. 71 6. 04     W B C en te ra -1 .7 5 -6 .9 -5 .1 5 -1 1. 56               W B S ou th -0 .7 3 -2 .8 6 -0 .7 2 -1 .7 9   1. 16 9. 29 1. 16 9. 29     G S N or th 2. 22 8. 54 2. 58 6. 93   0. 19 1. 61 0. 19 1. 61   2. 10 7. 95 2. 28 6. 22   G S C en te rb           1. 30 5. 27 1. 30 5. 27             G S So ut h 2. 15 7. 92 2. 45 6. 36   1. 91 5. 08 1. 91 5. 08     2. 09 7. 6 2. 32 6. 09     ru ra l 0. 02 0. 18 0. 43 1. 67     -1 .5 9 -2 .9 7 -3 .2 1 -3 .9 7   ca m p 1. 42 7. 28 2. 70 9. 20   1. 59 5. 37 2. 20 5. 81 co ns ta nt 3. 25 7. 73 -0 .7 2 -0 .9 6     2. 63 3. 33 -1 .3 6 -1 .2 4 R 2 0. 31       0. 23 0. 23       a W B C en te r, w h er e R am al la h a n d E as t Je ru sa le m a re lo ca te d , i s as su m ed a s re fe re n ce in t h e W es t B an k m o d el ; b G S C en te r, w h er e G az a C it y is lo ca te d , i s as su m ed a s re fe re n ce in b o th t h e Pa le st in e an d G az a St ri p m o d el s. 49The impact of assistance on poverty and food security in a fragile and protracted-crisis context Ta b le 1 3. F C S re g re ss io n m o d el s. V ar ia bl es   Pa le st in e W es t B an k G az a St ri p Po ol ed O LS Fi xe d eff ec ts Po ol ed O LS Fi xe d eff ec ts Po ol ed O LS Fi xe d eff ec ts C oe f. St ud en t’s t C oe f. St ud en t’s t C oe f. St ud en t’s t C oe f. St ud en t’s t C oe f. St ud en t’s t C oe f. St ud en t’s t m as s -0 .8 4 -9 .6 4 -0 .5 8 -6 .2 2 -0 .8 0 -6 .3 6 -0 .7 0 -6 .2 2 -1 .1 4 -8 .5 4 -1 .0 4 -6 .2 2 yd um -0 .0 1 -0 .0 4 -0 .2 7 -0 .7 8 0. 35 0. 98 -0 .1 0 -0 .2 8 -1 .9 3 -1 .9 8 -3 .7 8 -3 .2 8 m as sy -0 .2 6 -2 .2 8 -0 .2 2 -1 .6 8 0. 21 1. 19 0. 58 2. 63 -0 .1 3 -0 .6 4 0. 27 1. 01 lh si ze 3. 29 8. 79 4. 51 9. 22 3. 34 7. 37 4. 42 7. 57 3. 84 5. 81 5. 11 5. 76 de p_ ra tio 0. 27 2. 44 0. 26 1. 82 0. 29 2. 21 0. 29 1. 76 0. 09 0. 38 0. 10 0. 33 ra t_ em p 7. 68 11 .0 3 8. 27 9. 85 6. 96 8. 57 8. 51 8. 48 9. 79 7. 31 8. 03 5. 3 em pl oy ed 1. 25 2. 98 -0 .4 6 -2 .1 1. 39 2. 74 -0 .4 7 -1 .8 4 0. 62 0. 85 -0 .4 3 -1 .0 2 ag eh ea d 0. 07 6. 1 0. 05 4. 31 0. 09 4. 51 re fh ea d -0 .4 5 -1 .4 4     0. 61 1. 58 -2 .5 9 -4 .6 9 fe m he ad -0 .8 6 -1 .6 2     -0 .7 5 -1 .1 9 -1 .0 2 -1 .0 3 hi gh _e d 3. 28 11 .1 6     2. 83 8. 07 3. 46 6. 42 W B N or th -0 .6 7 -0 .9 3     -0 .7 7 -1 .8 9 W B C en te ra 0. 00 0             W B S ou th -9 .8 8 -1 3. 15     -1 0. 43 -2 2. 66 G S N or th -5 .7 6 -7 .9 5     -4 .8 1 -6 .5 4 G S C en te rb             G S So ut h -7 .4 9 -9 .7 1     -6 .7 0 -8 .6 ru ra l -1 .2 6 -3 .4 2     -1 .9 0 -4 .9 5 4. 24 3. 15 ca m p -1 .6 9 -3 .0 7     -4 .4 2 -5 .7 5 0. 67 0. 86 co ns ta nt 67 .2 7 53 .9 5     66 .2 8 49 .8 7 68 .5 7 34 .3 8 R 2 0. 13   0. 06 0. 13   0. 03 0. 13   0. 06   F te st o f fi xe d eff ec t  1. 28 p= 0. 0 1. 36 p= 0. 0   1. 28 p= 0. 0 a W B C en te r, w h er e R am al la h a n d E as t Je ru sa le m a re lo ca te d , i s as su m ed a s re fe re n ce in t h e W es t B an k m o d el ; b G S C en te r, w h er e G az a C it y is lo ca te d , i s as su m ed a s re fe re n ce in b o th t h e Pa le st in e an d G az a St ri p m o d el s. 50 D. Romano et alii The FCS results are quite different. According to OLS estimates (first column of Table 13), the quality of food consumption in Palestine seems to be negatively affected by the intensity of assistance.23 However, in the fixed effects model, the interaction parameter is not significant. All variables whose coefficients are statistically significant show the same signs as in the poverty index models except for two cases: the dependency ratio and the household size. They both have a positive effect on FCS, possibly because a larger num- ber of household members includes a sizeable share of children and elders calling for par- ticularly dietary requirements and/or making the household more eligible for food aid targeting. Regional dummies are all negative vis-à-vis Central Gaza except for the North and Central West Bank. The latter two regions show non-significant coefficients, possibly explained by higher population density and more urban nature. The West Bank and Gaza Strip models provide quite a different picture when consid- ering the fixed effect model. The impact of assistance on FCS is positive and significant in the West Bank but it is not significant in the Gaza Strip. This may depend on the nature of the outcome variable. A higher FCS presupposes the availability and physical accessibility of a variety of food, a condition that may not have held in Gaza Strip because of the open armed conflict and strict blockade that occurred in 2014. Keeping in mind that under these very specific conditions food security was pursued primarily through humanitarian assistance, we have to consider that in-kind food aid is based on food baskets containing only basic foodstuffs such as wheat flour, rice, pulses and vegetable oil. Therefore, in order to assess the impact of assistance on FCS via in-kind food aid, we disentangled the overall FCS in three additive components24 and estimated the impact model per each FCS component (Table 14). Doing so resulted in a slightly different picture. The intensity of assistance showed a positive impact of the two components provided via in-kind food assistance. The first component, which includes cereals, tubers, pulses, fruits and vegetables, is positive though significant only at p=90%. The second component, which includes oil and sugar, has a positive and significant impact at p=95%. Conversely, the component not included in the food aid basket, i.e. the meat and milk component, was not significant. This may be attrib- uted in part to the nature of in-kind food assistance constituted of cereals, pulses and veg- etable oil during the war in Gaza and in part to the low-income elasticity of these food categories as a source of low-cost calories and proteins. The less significant relationship found with reference to the first components can be explained by the dramatic drop in the availability of fruit and vegetables in the Gaza Strip as a result of the war.25 This drop was only partially compensated by the in-kind food assistance of cereals and pulses. In conclu- sion, food security was ensured more in terms of the quantity of food provided than the reported that in practice up to 300 metres from the perimeter fence is considered by most farmers as a “no-go” area and up to 1,000 metres a “high risk” area” (OCHA, 2018: 5). This area is where most military operations take place. 23 Higher FCS scores means in fact higher food quality as it measures food security in term of diet diversifica- tion. 24 The three components and the relevant FCS weights are the following: fruits, vegetables, cereals, tubers and pulses (weights from 1 to 3); milk and meats (weight equal to 4); oil, sugar and others (weight equal to 0.5). 25 Commercial food imports to the Gaza Strip cover a significant share of Gazan food needs. They stopped almost completely in the second half of 2014 because of the war and were partially offset by humanitarian imports providing food aid (Latino and Flämig, 2017). 51The impact of assistance on poverty and food security in a fragile and protracted-crisis context quality of diet during the war and following the conclusion of the hostility, at the height of the humanitarian crisis when interventions were primarily a matter of saving lives. 7. Conclusions This paper contributes to the scanty literature on the impact of humanitarian assis- tance interventions and outcomes (Clarke et al., 2014). It aims to answer a question that, to the best of our knowledge, has yet to be addressed: does assistance – broadly defined as any type of in-kind or cash transfer – improve the well-being of Palestinian households? To do so, we apply advanced econometric techniques and impact evaluation approaches widely advocated in the debate on aid effectiveness (cf. section 2.2). Specifically, we cou- pled the classical counterfactual framework of impact evaluation analysis with fixed effect econometric modelling using a difference-in-difference approach. This allowed us to treat sample selection bias. We also instrumented the fixed effect model to get rid of endogene- ity where needed, such as in poverty models. The main results are in line with existing literature (Ruel et al., 2013). Assistance is indeed crucial to support the standards of living of Palestinians: both poverty and food insecurity would have been much higher without the massive assistance provided by the international community to Palestine. This result supports similar conclusions attained by recent studies on contexts marked by violent conflicts and food insecurity crises (Doocy and Tappis, 2016; Mercier et al., 2017; Trachant et al., 2018). We confirmed the key role played by assistance, specifically food aid, extending the evidence to a protracted crisis context such as Palestine. The first policy implication is therefore that the international community should not keep disengaging from supporting Palestinian households. Over the last decade, over- all assistance to Palestine shrank by two thirds since 2008. The international commu- nity should be aware that if assistance continues to diminish, the severely negative con- sequences on the ground will affect the wellbeing of these households. More generally, the positive impact of assistance on poverty reduction and food security established in Table 14. FCS components fixed effect regression models, Gaza Strip Total Cereals, pulses, vegetables & fruit Meat & milk Oil & sugar Coef. Student ‘s t Coef Student’s t Coef Student’s t Coef Student’s t mass -1.04 -5.44 -0.20 -3.31 -0.83 -5.01 -0.01 -0.97 ydum -3.78 -3.28 0.23 0.62 -3.37 -3.46 -0.63 -8.12 massy 0.27 1.01 0.14 1.72 0.09 0.39 0.03 2.01 lhsize 5.11 5.76 1.39 5.18 3.37 4.27 0.35 5.77 dep_ratio 0.10 0.33 -0.06 -0.71 0.23 0.88 -0.07 -3.39 rat_emp 8.03 5.30 -0.15 -0.31 8.14 6.21 0.05 0.51 employed -0.43 -1.02 0.24 1.87 -0.66 -1.75 -0.02 -0.53 R2 0.07 0.02 0.07 0.06 52 D. Romano et alii this paper encourages renewed investment and further effort in enhancing aid effective- ness through better coordination of implementing actors and better design, targeting and delivery of assistance to the Palestinian people. It is important to keep in mind that the average positive impact of assistance hides a lot of heterogeneity with marked differences on each outcome dimension (poverty, quan- tity of food consumed, diet diversity) and region (West Bank or Gaza Strip). In the case of poverty reduction, there is a clear positive impact of intensity of assistance for both Palestine as a whole and the West Bank. However, this relationship is not significant for the Gaza Strip, probably because of the July-August 2014 war that could have blurred the causal relationship between assistance and poverty reduction. Assistance has a positive and significant impact on the amount of food consumed (proxied by HFIAS) in both regions, though the impact is much larger in the Gaza Strip than in the West Bank. This is thanks to massive in-kind food aid, food vouchers and cash interventions during and after the 2014 war that helped keep levels of food consumption at an acceptable level and restore household resilience (Brück et al., 2018). In the case of diet diversity (proxied by the FCS), there is no significant impact of assistance for Pales- tine as a whole. The impact is however significantly positive for the West Bank but not for the Gaza Strip. When disentangling this last result according to main diet components, we see that the two components included in the food basket provided to households in need – cereals and pulses, and oil and sugar – have positively affected Gazan households. This is true despite the fact that in-kind food aid was only partially able to compensate for the dramatic drop in the availability of fruit and vegetables imports during and after the 2014 military escalation. A second policy implication therefore relates to the importance of the composi- tion of food baskets provided to a population in need in order to ensure a balanced diet (Webb et al., 2014). This issue was raised in recent worldwide debates, specifically in Pal- estine where the food basket provided by UNRWA (OCHA, 2016) and by WFP (2017a and 2017b), the two most important implementing agencies, recently changed in order to provide more fortified and balanced food baskets. Careful consideration of the composi- tion of food baskets is extremely important, especially when considering long-term con- sequences of a balanced diet to targeted households with children (Alderman et al., 2006). Our study presents some limits. Understanding why assistance determined the above- mentioned outcomes would require more detailed information as well as an informa- tion-eliciting tool different from the one used by the SEFSec. Indeed, the SEFSec dataset, although quite informative on quantitative aspects of assistance to Palestinian households, is not able to open the black box of mechanisms that lead to these outcomes. Nor was it possible to analyze the effectiveness of different forms and sources of assistance, which affect the logics of intervention in a different manner. Addressing these topics would have required a larger and more detailed database supplemented by qualitative information, which we did not have. Nevertheless, the SEFSec dataset may be further exploited to shed light on issues such as the spatial distribution of assistance. The dataset could even be used to conduct a finer analysis of the impact of different types of assistance on food security as soon as the third wave (carried out in late 2018) data is made available. Methodological speaking, a possible future improvement to consider would be to model the different impact of assistance on 53The impact of assistance on poverty and food security in a fragile and protracted-crisis context asset accumulation/decumulation or even on household expenditure, provided the data is of adequate quality. 8. References Acemoglu, D., Autor, D.H., and Lyle, D. (2004). Women, War, and Wages: The Effect of Female Labor Supply on the Wage Structure at Mid-century. Journal of Political Economy 112(3): 497-551. Acemoglu, D., Johnson, S., and Robinson, J.A. (2005). “Institutions as the Fundamental Cause of Long-Run Growth”. In Aghion, P., and Durlauf, S. (eds.) Handbook of Eco- nomic Growth. Amsterdam: Elsevier Science, North-Holland, 385-472. Addison, T. (2000). “Aid and conflict”. In Tarp, F., (ed.) Foreign Aid and Development. Les- sons Learnt and Directions for the Future. London and New York: Routledge, 305-317. Ahmed, A.U., Quisumbing, A.R., Nasreen, M., Hoddinott, J.F. and Bryan, E. (2009). Com- paring Food and Cash Transfers to the Ultra Poor in Bangladesh. IFPRI Research Monograph 163. Washington, DC: International Food Policy Research Institute. http://www.ifpri.org/publication/comparing-food-and-cash-transfers-ultra-poor- bangladesh. Alderman, H., Hoddinott, J., and Kinsey, B. (2006). Long Term Consequences of Early Childhood Malnutrition. Oxford Economic Papers 58(3): 450–474. Andrimihaja, N.A., Cinyabuguma, M., and Devarajan, S. (2011). Avoiding the Fragil- ity Trap in Africa. World Bank Working Paper 5884. Washington, D.C.: The World Bank. https://elibrary.worldbank.org/doi/abs/10.1596/1813-9450-5884. Angrist, J.D., and Pischke, J.S. (2008). Mostly Harmless Econometrics: An Empiricist’s Com- panion. Princeton, NJ: Princeton University Press. Atamanov, A., and Palaniswamy, N. (2018). West Bank and Gaza Poverty and Shared Prosperity Diagnostic 2011-2017. Washington, D.C.: The World Bank. August 14, 2018. http://documents.worldbank.org/curated/en/985801536125679274/West- Bank-and-Gaza-Poverty-and-Shared-Prosperity-Diagnostic-2011-2017. Baliamoune-Lutz, M. and McGillivray, M. (2008). State Fragility – Concept and Measure- ment. WIDER Research Paper No. 2008/44. Helsinki: World Institute for Develop- ment Economics Research. Baltagi, B.H. (2005). Econometric Analysis of Panel Data (third ed.). Chichester: John Wiley & Sons. Banerjee, A. (2007). Making Aid Work: How to Fight Global Poverty Effectively. Cambridge and London: MIT Press. Banerjee, A., and Duflo, E. (2011). Poor Economics: A Radical Rethinking of the Way to Fight Global Poverty. New York: Public Affairs. Bauer. P., and Yamey, B. (1982). Foreign Aid: What Is at Stake? Public Interest, Summer (68): 53-69. Baum, C.F., Schaffer, M.E., and Stillman, S. (2007). Enhanced Routines for Instrumental Variables/GMM Estimation and Testing. Stata Journal 7(4): 465-506. Birchler, K., and Michaelowa, K. (2016). Making Aid Work for Education in Developing Countries: An Analysis of Aid Effectiveness for Primary Education Coverage and Quality. International Journal of Educational Development 48(1): 37-52. 54 D. Romano et alii Boone, P. (1995). Politics and the Effectiveness of Foreign Aid. European Economic Review 40(2): 289–329. Bourguignon, F., and Sundberg, M. (2007). Aid Effectiveness: Opening the Black Box. American Economic Review 97(2): 316-321. Brück, T., d’Errico, M., and Pietrelli, R. (2018). The Effects of Violent Conflict on House- hold Resilience and Food Security: Evidence from the 2014 Gaza Conflict. World Development (2018), https://doi.org/10.1016/j.worlddev.2018.05.008. Burnside, C., and Dollar, D. (2000). Aid, Policies, and Growth. American Economic Review 90(4): 847-868. Cameron, D.B., Brown, A.N. Mishra, A., Picon, M., Esper, H., Calvo, F., and Peterson, K. (2015). Evidence for Peacebuilding: Evidence Gap Map. 3ie Evidence Gap Report 1, April 2015. New Delhi: International Initiative for Impact Evaluation (3ie). Cameron, D.B., Mishra, A., and Brown, A.N. (2016). The Growth of Impact Evaluation for International Development: How Much Have We Learned? Journal of Development Effectiveness 8(1): 1-21. Card, D. (1992), Using Regional Variation in Wages to Measure the Effects of the Federal Minimum Wage. Industrial and Labor Relations Review 46(1): 22-37. Carletto, C., Zezza, A., and Banerjee, R., (2013). Towards Better Measurement of House- hold Food Security: Harmonizing Indicators and The Role of Household Surveys. Global Food Security 2(1): 30-40. Cassens, R., and Ass. (1986). Does Aid Work? Report to an Intergovernmental Task Force. Oxford: Clarendon. Chandy, L. (2011). Ten Years of Fragile States: What Have We Learned? Brookings Institu- tion Global Views Policy Paper. Washington, D.C.: The Brookings Institution. htt- ps://www.brookings.edu/research/ten-years-of-fragile-states-what-have-we-learned/. Chandy, L., Seidel, B., and Zhang, C. (2016). Aid Effectiveness in Fragile States. How Bad Is It and How Can It Improve? Brooke Shearer Series no. 5. Washington, D.C.: The Brookings Institution. December 2016. https://www.brookings.edu/research/aid- effectiveness-in-fragile-states/. Chenery, H.B., and Bruno, M. (1962). Development Alternatives in an Open Economy: The Case of Israel. Economic Journal 72(285): 79-103. Chenery, H.B., and Strout, A.M. (1966). Foreign Assistance and Economic Development. American Economic Review 56(4): 679-733. Clarke, M., Allen, C., Archer, F., Wong, D., Eriksson, A., and Puri, J. (2014). What Evi- dence Is Available and What Is Required, in Humanitarian Assistance? 3ie Scoping Paper 1, December 2014. New Delhi: International Initiative for Impact Evaluation (3ie). Clemens, M., Radelet, S. and Bhavnani, R. (2004). “Counting Chickens When They Hatch: The Short-Term Effect of Aid on Growth”. Working Paper 44, Washington D.C.: Center for Global Development. http://dx.doi.org/10.2139/ssrn.567241. Coates, J., Swindale, A., and Bilinsky, P. (2007). Household Food Insecurity Access Scale (HFIAS) for Measurement of Household Food Access: Indicator Guide (v. 3). Wash- ington, D.C.: Food and Nutrition Technical Assistance Project, Academy for Edu- cational Development, August 2007. http://www.fantaproject.org/sites/default/files/ resources/HFIAS_ENG_v3_Aug07.pdf. 55The impact of assistance on poverty and food security in a fragile and protracted-crisis context Collier, P. (2007). The Bottom Billion: Why the Poorest Countries are Failing and What Can Be Done About It. Oxford: Oxford University Press. Collier, P., and Dollar, D. (2001). Can the World Cut Poverty in Half ? How Policy Reform and Effective Aid Can Meet International Development Goals. World Development 29(11): 1787-1802. Collier, P., and Dollar, D. (2002). Aid Allocation and Poverty Reduction. European Eco- nomic Review 46(8): 1475–500. Culbert, V. (2017). Social Development, Food and Cash Assistance in Palestine. Overview of Current Practices. November 2017. Dalgaard, C.G., and Hansen, H. (2001). On Aid, Growth and Good Policies. Journal of Development Studies 37(6): 17–41. Dalgaard, C.G., Hansen, H., and Tarp, F. (2004). On the Empirics of Foreign Aid and Growth. Economic Journal 114 (June): F191-F216. Deaton, A. (2013). The Great Escape: Health, Wealth, and the Origins of Inequality. Prince- ton, NJ: Princeton University Press. Dethier, J.J. (2008). “Aid Effectiveness: What Can We Know? What Should We Know? What May We Hope?” Washington, D.C.: The World Bank. Diakonia (2018). The Humanitarian - Development Divide: A False Dichotomy? The International Law Framework for Humanitarian and Development Assistance in a Context of Protracted Occupation. Diakonia, International Humanitarian Law Resource Centre, August 2018. https://www.diakonia.se/en/IHL/News-List/the- humanitarian-development-divide-a-false-dichotomy/. Dollar, D. and Levin, V. (2006). The Increasing Selectivity of Foreign Aid, 1984-2003. World Development 34(12): 2034-2046. Domar, E. (1957). Essays in the Theory of Economic Growth. Oxford: Oxford University Press. Doocy, S., and Tappis, H. (2016). Cash-based Approaches in Humanitarian Emergencies: A Systematic Review. 3ie Systeatic Review Report 28, April 2016. London: Interna- tional Initiative for Impact Evaluation (3ie). Doucouliagos, H., and Paldam, M. (2009). The Aid Effectiveness Literature: The Sad Results of 40 Years of Research. Journal of Economic Surveys 23(3): 433-461. Easterly, W. (2006). The White Man’s Burden. New York: Penguin Press. Easterly, W., Levine, R., and Roodman, D. (2004). New Data, New Doubts: A Comment on Burnside and Dollar’s “Aid, Policies, and Growth” (2000). American Economic Review 94(3): 774-780. Elkhafif, M., Misyef, M., and Elagraa, M. (2014). Palestinian Fiscal Revenue Leakage to Israel under the Paris Protocol on Economic Relations. New York and Geneva: Unit- ed Nations Conference on Trade and Development. EU Commission (2009). European Report on Development 2009. Overcoming Fragility in Africa. Robert Schuman Centre for Advanced Studies, European University Institute, San Domenico di Fiesole. http://erd.eui.eu/media/fullreport/ERD%202009_EN_ LowRes.pdf. EU Commission (2018). Towards a Democratic and Accountable Palestinian State. Euro- pean Joint Programming Strategy in Support of Palestine 2017-2022. Bruxelles: European External Action Service. http://eueuropaeeas.fpfis.slb.ec.europa.eu:8084/ headquarters/headquarters-homepage/25035/eu-joint-programming-palestine_en. 56 D. Romano et alii FAO – Food and Agriculture Organization of the UN, UNRWA – United Nations Relief and Works Agency for Palestine Refugees in the Near East, WFP – World Food Pro- gramme and WHO – World Health Organization (2013). Socio-Economic & Food Security Survey 2012. Jerusalem: FAO. FAO – Food and Agriculture Organization of the UN, IFAD – International Fund for Agricultural Development, UNICEF – United Nations International Children’s Fund, WFP – World Food Programme and WHO – World Health Organization (2017). The State of Food Security and Nutrition in the World 2017. Building Resil- ience for Peace and Food Security. Rome: FAO. http://www.fao.org/policy-support/ resources/resources-details/en/c/1037641/. Feeny, S. and McGillivray, M. (2009). Aid Allocation to Fragile States: Absorptive Capacity Constraints. Journal of International Development 21(5): 618-632. Fielding, D. and Mavrotas, G. (2008). Aid Volatility and Donor-Recipient Characteristics in Difficult Partnership Countries. Economica 75(299): 481-494. FSS-PCBS (2016). Socio-Economic & Food Security Survey 2014. Ramallah: Food Security Sector and Palestinian Central Bureau of Statistics. http://fscluster.org/sites/default/ files/documents/sefsec2014_report_all_web.pdf. Gomanee, K., Girma, S., and Morrissey, O. (2005). Aid and Growth in Sub-Saharan Afri- ca: Accounting for Transmission Mechanisms. Journal of International Development 17(8): 1055-1075. Greene, W. (2004). The Behaviour of the Maximum Likelihood Estimator of Limited Dependent Variable Models in the Presence of Fixed Effects. Econometrics Journal 7: 98-119. Guisselquist, R.M. (2015). Good Aid in Hard Places: Learning from ‘Successful’ Interven- tions in Fragile Situations. International Peacekeeping 22(4): 283-301. Hansen, H., and Tarp, F. (2001). Aid and Growth Regressions. Journal of Development Economics 64(2): 547-570. Hever, S. (2010). The Political Economy of Israel’s Occupation: Repression Beyond Exploita- tion. London: Pluto Press. Hever, S. (2015). How Much International Aid to Palestinians Ends Up in the Israeli Economy? Aid Watch, September 2015. http://www.aidwatch.ps/sites/default/ files/resource-field_media/InternationalAidToPalestiniansFeedsTheIsraeliEcono- my.pdf. Hirschman, A.O. (1967). Development Projects Observed. Washington, D.C.: The Brooking Institution. Honoré, B.E. (1992). Trimmed Lad and Least Squares Estimation of Truncated and Cen- sored Regression Models with Fixed Effects. Econometrica 60 (3): 533-565. Humphreys, M. (2015). Reflections on the Ethics of Social Experimentation. Journal of Globalization and Development 6(1): 87-112. Ikpe, E. (2007). Challenging the Discourse on Fragile States. Conflict Security & Develop- ment 7(1): 85-124. IFPRI – International Food Policy Research Institute (2006). Review and validation of die- tary diversity, food frequency and other proxy indicators of household food security. Washington, D.C.: World Food Programme, Vulnerability Analysis and Mapping Branch. July 2006. 57The impact of assistance on poverty and food security in a fragile and protracted-crisis context IMF – International Monetary Fund (2018). West Bank and Gaza. Report to The Ad Hoc Liaison Committee. September 6 2018. http://www.imf.org/wbg. Ishihara, Y. (2012). Identifying Aid Effectiveness Challenges in Fragile and Conflict- Affected States. Policy Research Working Paper No. 6037. Washington, D.C.: The World Bank. Kanafani, N., and Al-Botmeh, S. (2008). The Political Economy of Food Aid to Palestine. The Economics of Peace and Security Journal 3(2): 39-48. Kanbur, R. (2000). “Aid, Conditionality and Debt in Africa.” In Tarp, F., (ed.) Foreign Aid and Development. Lessons Learnt and Directions for the Future. London and New York: Routledge, 409-22. Kaplan, S. (2008). Fixing Fragile States – A New Paradigm for Development. Westport, CT: Praeger Security International. Karlan, D., and Appel, J. (2012). More Than Good Intentions: Improving the Ways the World’s Poor Borrow, Save, Farm, Learn, and Stay Healthy. New York: Plume. Khandker, S.R., Koolwal, G.B., and Samad, H.A. (2010). Handbook on Impact Evaluation. Quantitative Methods and Practices. Washington, D.C.: The World Bank. Knueppel, D., Demment, M., and Kaiser, L. (2010). Validation of the Household Food Insecurity Access Scale in Rural Tanzania. Public Health Nutrition 13(3): 360-367. Krueger, A.O. (1986). Aid in the Development Process. World Bank Research Observer 1(1): 57-78. Lal, D. (1972). The Foreign-Exchange Bottleneck Revisited: A Geometric Note. Economic Development and Cultural Change 20(4): 722-730. Langworthy, M., Smith, L.C., and Sagara, B. (2014). Review of Palestine SEFSec Food Security Analysis Methodology. Report #1. Tucson, AZ: TANGO International, Inc. 18 March 2014. Latino, L., and Flämig, T. (2017). Market Assessment in the Gaza Strip. Is the market of the besieged enclave conducive to a large CBT intervention? Rome: World Food Programme. McGillivray, M. and Feeny, S. (2008). Aid and Growth in Fragile States. WIDER Discus- sion Paper No. 2008/03. Helsinki: World Institute for Development Economics Research. McGillivray, M., Feeny, S., Hermes, N., and Lensink, R. (2006). Controversies over the Impact of Development Aid: It Works, It Doesn’t, It Might, but that Depends. Jour- nal of International Development 18(7): 1031-1050. Manley, J., Gitter, S., and Slavchevska, V. (2012). How Effective Are Cash Transfer Pro- grammes at Improving Nutritional Status? A Rapid Evidence Assessment of Pro- grammes’ Effects on Anthropometric Outcomes. London: EPPI-Centre, Social Sci- ence Research Unit, Institute of Education, University of London. Masino, S., and Niño-Zarazúa, M. (2016). What Works to Improve The Quality of Student Learning in Developing Countries? International Journal of Educational Develop- ment 48(1): 53-65. Mavrotas, G. (2015). “The Macroeconomic Impact of Aid in Recipient Countries: Old Wine in New Bottles?”. In Arvin, M.B., and Lew, B. (Eds.), Handbook on the Eco- nomics of Foreign Aid. Cheltenham: Edward Elgar. Pp. 215-231. Mekasha, T.J., and Tarp, F. (2013). Aid and Growth: What Meta-Analysis Reveals. Journal of Development Studies 49(4): 564-583. 58 D. Romano et alii Melgar-Quinonez, H.R., Zubieta, A.C., MkNelly, B., Nteziyaremye, A., Gerardo, M.F., Dunford, C. (2006). Household Food Insecurity and Food Expenditure in Bolivia, Burkina Faso, and the Philippines. Journal of Nutrition 136(5): 1431S–1437S. Mercier, M., Ngenzebuke, R.L., and Verwimp, H.P. (2017). Violence Exposure and Depri- vation: Evidence from the Burundi Civil War. Working Papers DT/2017/14, DIAL (Développement, Institutions et Mondialisation). MoSD – Ministry of Social Development (2017). Social Development Sector Strategy (2017-2022). Ramallah: Ministry of Social Development, State of Palestine. February 2017. Mosley, P. (1987). Foreign Aid: Its Defense and Reform. Lexington, KY: University Press of Kentucky. Moyo, D. (2009). Dead Aid: Why Aid Is Not Working and How There is a Better Way for Africa. New York: Farrar, Straus and Giroux. Murshed, S.M. (2002). Conflict, Civil War and Underdevelopment: An Introduction. Jour- nal of Peace Research 39(4): 387-393. Naudé, W.A., Santos-Paulino, A.U., and McGillivray, M. (2011). Fragile States: Causes, Costs and Responses. Oxford: Oxford University Press. OCHA – United Nations Office for the Coordination of Humanitarian Affairs (2016). Revised Food Baskets Aim to Improve Nutrition Among Food Insecure Refugees in Gaza. The Monthly Humanitarian Bullettin, February 2016. https://www.ochaopt. org/content/revised-food-baskets-aim-improve-nutrition-among-food-insecure-ref- ugees-gaza. OCHA – United Nations Office for the Coordination of Humanitarian Affairs (2017a). 2018-2020 Humanitarian Response Strategy. Humanitarian Response Plan January- December 2018. Jerusalem: OCHA. December 2017. www.ochaopt.org. OCHA – United Nations Office for the Coordination of Humanitarian Affairs (2017b). New Way of Working. New York: OCHA, Development and Studies Branch. https:// www.unocha.org/es/themes/humanitarian-development-nexus. OCHA – United Nations Office for the Coordination of Humanitarian Affairs (2018). Humanitarian Needs Overview 2019. Jerusalem: OCHA. December 2018. www. ochaopt.org. OECD – Organisation for Economic Co-operation and Development (2007). Principles for Good International Engagement in Fragile States and Situations. Paris: Organisa- tion for Economic Co-operation and Development/Development Assistance Com- mittee. OECD – Organisation for Economic Co-operation and Development (2015). States of Fragility 2015. Meeting Post-2015 Ambitions. Paris: OECD Organization for Eco- nomic Co-operation and Development. http://www.oecd.org/dac/states-of-fragility- 2015-9789264227699-en.htm. PCBS – Palestinian Central Bureau of Statistics (2018a). Levels of Living in Palestine, 2017. News. at http://www.pcbs.gov.ps/post.aspx?lang=en&ItemID=3115 Accessed on July 3rd 2018 PCBS – Palestinian Central Bureau of Statistics (2018b). SEFSec 2018 - Food Security Analysis Preliminary Results. https://fscluster.org/state-of-palestine/document/sef- sec-2018-food-security-analysis. 59The impact of assistance on poverty and food security in a fragile and protracted-crisis context Pritchett, L., and Sandefur, J. (2013). “Context Matters for Size: Why External Valid- ity Claims and Development Practice Don’t Mix”. Center for Global Development Working Paper 336, Washington, D.C.: Center for Global Development. Rajan, R., and Subramanian, A. (2005). “Aid and Growth: What Does the Cross-Country Evidence Really Show?” Washington, D.C.: IMF Working Paper 05-127. Riddell, A., and Niño-Zarazúa, M. (2016). The Effectiveness of Foreign Aid to Education. What Can Be Learned? International Journal of Educational Development 48(1): 23-36. Robinson, S., and Tarp, F. (2000). “Foreign Aid and Development”. In Tarp, F., (ed.) For- eign Aid and Development. Lessons Learnt and Directions for the Future. London and New York: Routledge. Pp. 1-10. Rosenstein-Rodan, P. (1943). Problems of Industrialization of Eastern and South-Eastern Europe. Economic Journal 53(210): 202-11. Ruel, M.T., Alderman, H., and Maternal and Child Nutrition Study Group (2013). Nutri- tion‐sensitive Interventions and Programmes: How Can They Help to Accelerate Pro- gress in Improving Maternal and Child Nutrition? The Lancet 382(9891): 536-551. Sachs, J. (2005). The End of Poverty: Economic Possibilities for Our Time. New York: Pen- guin Press. Savedoff, W.D., Levine, R. and Birdsall, N. (2006). When Will We Ever Learn? Improv- ing Lives through Impact Evaluation. Report of the Evaluation Gap Working Group, May 2006. Washington, D.C.: Center for Global Development. Smith, L.C. (2014). Review of Palestine SEFSec Food Security Analysis Methodology, Phase II. Tucson, AZ: TANGO International, Inc. December 2014. Sogge, D. (2002). Give and Take: What’s the Matter with Foreign Aid? London: Zed Books. Stewart, F. and Brown, G. (2009). Fragile States. Working Paper No. 51. Centre for Research on Inequality, Human Security and Ethnicity, Oxford, UK. Taghdisi-Rad, S. (2011). The Political Economy of Aid in Palestine: Relief from Conflict or Development Delayed? London: Routledge. Taghdisi-Rad, S. (2015). Political Economy of Aid in Conflict: An Analysis of Pre- and Post-Intifada Donor Behaviour in the Occupied Palestinian Territories. Stability: International Journal of Security & Development 4(1): 1-18. Thorbecke, E. (2000). “The Evolution of the Development Doctrine and the Role of For- eign Aid, 1950-2000.” In Tarp, F., and Hjertholm, P. (eds.). Foreign Aid and Develop- ment – Lessons Learnt and Directions for the Future. London and New York: Rout- ledge. Pp. 17-47 Trachant, J.P., Gelli, A., Bliznashka, L., Diallo, A. S., Sacko, M., Assima, A., Siegel, E., Auri- no, E., and Masset, E. (2018). The Impact of Food Assistance on Food Insecure Pop- ulations During Conflict: Evidence from A Quasi-Experiment in Mali. World Devel- opment (2018), https://doi.org/10.1016/j.worlddev.2018.01.027. UN – United Nations (2016). One Humanity: Shared Responsibility. Report of the Secre- tary-General for the World Humanitarian Summit. United Nations General Assem- bly, Seventieth session, Item 73 (a), A/70/709. Distr.: General, 2 February 2016. htt- ps://sgreport.worldhumanitariansummit.org/. UNCTAD – United Nations Conference on Trade and Development (2015). Report on UNC- TAD Assistance to the Palestinian People: Developments in the Economy of the Occu- 60 D. Romano et alii pied Palestinian Territory. TD/B/62/3. Geneva: UNCTAD. https://unctad.org/en/pages/ newsdetails.aspx?OriginalVersionID=1068&Sitemap_x0020_Taxonomy=UNCTAD%20 Home;#1540;#Assistance%20to%20the%20Palestinian%20People. UNCTAD – United Nations Conference on Trade and Development (2017). Report on UNCTAD Assistance to the Palestinian People: Developments in the Economy of the Occupied Palestinian Territory. TD/B/64/4. Geneva: UNCTAD. https://unctad. org/en/pages/PublicationWebflyer.aspx?publicationid=1845. UNCTAD – United Nations Conference on Trade and Development (2018). Report on UNCTAD Assistance to the Palestinian People: Developments in the Economy of the Occupied Palestinian Territory. TD/B/65(2)/3. Geneva: UNCTAD. https:// unctad.org/en/pages/newsdetails.aspx?OriginalVersionID=1846&Sitemap_x0020_ Taxonomy=UNCTAD%20Home;#1540;#Assistance%20to%20the%20Palestinian%20 People. UNDP – United Nations Development Programme (2018). Human Development Indices and Indicators 2018 Statistical Update. New York: United Nations Development Pro- gramme. http://hdr.undp.org/en/content/human-development-indices-indicators- 2018-statistical-update. UNEG – United Nations Evaluation Group (2018). The Humanitarian Development Nexus – What Do Evaluations Have To Say? Mapping and Synthesis of Evaluation. UNEG Working Paper, UNEG Humanitarian Evaluation Interest Group, February 2018. http://www.uneval.org/document/detail/2120. UNSCO – United Nations United Nations Special Coordinator for the Middle east Peace Process (2016). Common Country Analysis 2016. Leave No One Behind: A Per- spective on Vulnerability and Structural Disadvantage in Palestine. Jerusalem: UN Country Team, occupied Palestinian Territory. http://www.unsco.org/Documents/ Special/UNCT/CCA_Report_En.pdf. Webb, P., Boyd, E., de Pee, S., Lenters, L., Bloem, M., and Schultink, W. (2014). Nutrition in Emergencies: Do We Know What Works? Food Policy 49 (1): 33-40. White, H. (2010). A Contribution to Current Debates in Impact Evaluation. Evaluation 16(2): 153-64. WDI – World Development Indicators (2018). World Development Indicators. Wash- ington, DC: The World Bank. http://databank.worldbank.org/data/reports. aspx?source=world-development-indicators Accessed on 20th December 2018. WFP – World Food Programme (2008). Food Consumption Analysis: Calculation and Use of The Food Consumption Score in Food Security Analysis. Prepared by VAM unit HQ Rome, Version 1. Rome: World Food Programme. February 2008. WFP – World Food Programme (2017a). WFP State of Palestine. Country Brief 2017. Jerusalem: World Food Programme in Palestine. https://reliefweb.int/report/occu- pied-palestinian-territory/wfp-state-palestine-country-brief-december-2017. WFP – World Food Programme (2017b). Food Assistance for the Food-Insecure Popula- tions in the West Bank and Gaza Strip. Standard Project Report 2017. Jerusalem: World Food Programme in Palestine. https://www1.wfp.org/operations/200709- food-assistance-food-insecure-population-west-bank-and-gaza-strip. Wooldridge, J.M. (2013). Introductory Econometrics, Econometrics: A Modern Approach, Fifth Edition. Mason, OH: South-Western, Cengage Learning. 61The impact of assistance on poverty and food security in a fragile and protracted-crisis context World Bank (1998). Assessing Aid: What Works, What Doesn’t, and Why. Washington, D.C.: The World Bank. World Bank (2002). The Role and Effectiveness of Development Assistance: Lessons from World Bank Experience. Washington, D.C.: The World Bank. World Bank (2011). World Development Report 2011: Conflict, Security and Development. Washington, D.C.: The World Bank. http://siteresources.worldbank.org/INTWDRS/ Resources/WDR2011_Full_Text.pdf. World Bank (2017). Unlocking the Trade Potential of the Palestinian Economy. Immedi- ate Measures and a Long-Term Vision to Improve Palestinian Trade and Economic Outcomes. Report No: ACS22471. Washington, DC: The World Bank. World Bank (2019). Net ODA received per capita (current US$). World Bank Open Data. https://data.worldbank.org/indicator/DT.ODA.ODAT.PC.ZS?most_recent_value_ desc=true. Zoellick, R.B. (2008). Fragile States: Securing Development. Survival 50(6): 67-84. Zseleczky, L., and Yosef, S. (2014). Are Shocks Really Increasing? A Selective Review of the Global Frequency, Severity, Scope, and Impact of Five Types of Shocks. In S. Fan, R. Pandya-Lorch, and S. Yosef (Eds.), Resilience for Food and Nutrition Security. Washingtn D.C.: IFPRI, 9-17. http://www.ifpri.org/publication/resilience-food-and- nutrition-security. The future of Bio-Based and Applied Economics Daniele Moro1, Fabio Gaetano Santeramo2, Davide Viaggi3 How did farmers act? Ex-post validation of linear and positive mathematical programming approaches for farm-level models implemented in an agent-based agricultural sector model Gabriele Mack*, Ali Ferjani, Anke Möhring, Albert von Ow, Stefan Mann The impact of assistance on poverty and food security in a fragile and protracted-crisis context: the case of West Bank and Gaza Strip Donato Romanoa, Gianluca Stefania,*, Benedetto Rocchia, Ciro Fiorilloa Assessing Price Sensitivity of Forest Recreational Tourists in a Mountain Destination Gianluca Grilli1,2 Estimating a Dual Value Function as a Meta-Model of a Detailed Dynamic Mathematical Programming Model Claudia Seidel, Wolfgang Britz